Visualization and TensorFlow.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Import Library\n",
    "import library"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 206,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: tensorflow in d:\\programdata\\anaconda3\\lib\\site-packages\n",
      "Requirement already satisfied: tensorflow-tensorboard<0.2.0,>=0.1.0 in d:\\programdata\\anaconda3\\lib\\site-packages (from tensorflow)\n",
      "Requirement already satisfied: numpy>=1.11.0 in d:\\programdata\\anaconda3\\lib\\site-packages (from tensorflow)\n",
      "Requirement already satisfied: wheel>=0.26 in d:\\programdata\\anaconda3\\lib\\site-packages (from tensorflow)\n",
      "Requirement already satisfied: six>=1.10.0 in d:\\programdata\\anaconda3\\lib\\site-packages (from tensorflow)\n",
      "Requirement already satisfied: protobuf>=3.3.0 in d:\\programdata\\anaconda3\\lib\\site-packages (from tensorflow)\n",
      "Requirement already satisfied: bleach==1.5.0 in d:\\programdata\\anaconda3\\lib\\site-packages (from tensorflow-tensorboard<0.2.0,>=0.1.0->tensorflow)\n",
      "Requirement already satisfied: werkzeug>=0.11.10 in d:\\programdata\\anaconda3\\lib\\site-packages (from tensorflow-tensorboard<0.2.0,>=0.1.0->tensorflow)\n",
      "Requirement already satisfied: markdown>=2.6.8 in d:\\programdata\\anaconda3\\lib\\site-packages (from tensorflow-tensorboard<0.2.0,>=0.1.0->tensorflow)\n",
      "Requirement already satisfied: html5lib==0.9999999 in d:\\programdata\\anaconda3\\lib\\site-packages (from tensorflow-tensorboard<0.2.0,>=0.1.0->tensorflow)\n",
      "Requirement already satisfied: setuptools in d:\\programdata\\anaconda3\\lib\\site-packages (from protobuf>=3.3.0->tensorflow)\n"
     ]
    }
   ],
   "source": [
    "!pip install tensorflow\n",
    "import seaborn as sns\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import math\n",
    "import tensorflow as tf\n",
    "import csv\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 207,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "b'Hello, TensorFlow!'\n"
     ]
    }
   ],
   "source": [
    "hello = tf.constant('Hello, TensorFlow!')\n",
    "sess = tf.Session()\n",
    "print(sess.run(hello))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Import Training set and Test set of Titanic Survival"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 208,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "shape of train set : (891, 12)\n",
      "shape of test set: (418, 11)\n",
      "Die rate : 38.38 %\n",
      " ---- Null data in train set : ----\n",
      " PassengerId      0\n",
      "Survived         0\n",
      "Pclass           0\n",
      "Name             0\n",
      "Sex              0\n",
      "Age            177\n",
      "SibSp            0\n",
      "Parch            0\n",
      "Ticket           0\n",
      "Fare             0\n",
      "Cabin          687\n",
      "Embarked         2\n",
      "dtype: int64\n",
      " ---- Null data in test set : ----\n",
      " PassengerId      0\n",
      "Pclass           0\n",
      "Name             0\n",
      "Sex              0\n",
      "Age             86\n",
      "SibSp            0\n",
      "Parch            0\n",
      "Ticket           0\n",
      "Fare             1\n",
      "Cabin          327\n",
      "Embarked         0\n",
      "dtype: int64\n",
      "    PassengerId  Survived  Pclass  \\\n",
      "0             1         0       3   \n",
      "1             2         1       1   \n",
      "2             3         1       3   \n",
      "3             4         1       1   \n",
      "4             5         0       3   \n",
      "5             6         0       3   \n",
      "6             7         0       1   \n",
      "7             8         0       3   \n",
      "8             9         1       3   \n",
      "9            10         1       2   \n",
      "10           11         1       3   \n",
      "11           12         1       1   \n",
      "12           13         0       3   \n",
      "13           14         0       3   \n",
      "14           15         0       3   \n",
      "15           16         1       2   \n",
      "16           17         0       3   \n",
      "17           18         1       2   \n",
      "18           19         0       3   \n",
      "19           20         1       3   \n",
      "\n",
      "                                                 Name     Sex   Age  SibSp  \\\n",
      "0                             Braund, Mr. Owen Harris    male  22.0      1   \n",
      "1   Cumings, Mrs. John Bradley (Florence Briggs Th...  female  38.0      1   \n",
      "2                              Heikkinen, Miss. Laina  female  26.0      0   \n",
      "3        Futrelle, Mrs. Jacques Heath (Lily May Peel)  female  35.0      1   \n",
      "4                            Allen, Mr. William Henry    male  35.0      0   \n",
      "5                                    Moran, Mr. James    male   NaN      0   \n",
      "6                             McCarthy, Mr. Timothy J    male  54.0      0   \n",
      "7                      Palsson, Master. Gosta Leonard    male   2.0      3   \n",
      "8   Johnson, Mrs. Oscar W (Elisabeth Vilhelmina Berg)  female  27.0      0   \n",
      "9                 Nasser, Mrs. Nicholas (Adele Achem)  female  14.0      1   \n",
      "10                    Sandstrom, Miss. Marguerite Rut  female   4.0      1   \n",
      "11                           Bonnell, Miss. Elizabeth  female  58.0      0   \n",
      "12                     Saundercock, Mr. William Henry    male  20.0      0   \n",
      "13                        Andersson, Mr. Anders Johan    male  39.0      1   \n",
      "14               Vestrom, Miss. Hulda Amanda Adolfina  female  14.0      0   \n",
      "15                   Hewlett, Mrs. (Mary D Kingcome)   female  55.0      0   \n",
      "16                               Rice, Master. Eugene    male   2.0      4   \n",
      "17                       Williams, Mr. Charles Eugene    male   NaN      0   \n",
      "18  Vander Planke, Mrs. Julius (Emelia Maria Vande...  female  31.0      1   \n",
      "19                            Masselmani, Mrs. Fatima  female   NaN      0   \n",
      "\n",
      "    Parch            Ticket     Fare Cabin Embarked  \n",
      "0       0         A/5 21171   7.2500   NaN        S  \n",
      "1       0          PC 17599  71.2833   C85        C  \n",
      "2       0  STON/O2. 3101282   7.9250   NaN        S  \n",
      "3       0            113803  53.1000  C123        S  \n",
      "4       0            373450   8.0500   NaN        S  \n",
      "5       0            330877   8.4583   NaN        Q  \n",
      "6       0             17463  51.8625   E46        S  \n",
      "7       1            349909  21.0750   NaN        S  \n",
      "8       2            347742  11.1333   NaN        S  \n",
      "9       0            237736  30.0708   NaN        C  \n",
      "10      1           PP 9549  16.7000    G6        S  \n",
      "11      0            113783  26.5500  C103        S  \n",
      "12      0         A/5. 2151   8.0500   NaN        S  \n",
      "13      5            347082  31.2750   NaN        S  \n",
      "14      0            350406   7.8542   NaN        S  \n",
      "15      0            248706  16.0000   NaN        S  \n",
      "16      1            382652  29.1250   NaN        Q  \n",
      "17      0            244373  13.0000   NaN        S  \n",
      "18      0            345763  18.0000   NaN        S  \n",
      "19      0              2649   7.2250   NaN        C  \n"
     ]
    }
   ],
   "source": [
    "whole_train_set = pd.read_csv('train.csv')\n",
    "whole_test_set = pd.read_csv('test.csv')\n",
    "\n",
    "print('shape of train set :',whole_train_set.shape)\n",
    "print('shape of test set:', whole_test_set.shape)\n",
    "\n",
    "print('Die rate :', round(np.mean(whole_train_set['Survived'])*100,2), '%')\n",
    "\n",
    "print(' ---- Null data in train set : ----\\n', whole_train_set.isnull().sum())\n",
    "print(' ---- Null data in test set : ----\\n', whole_test_set.isnull().sum())\n",
    "\n",
    "print(whole_train_set.loc[range(20)])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Cleanse Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 209,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "     Survived  Pclass     Sex   Age  SibSp  Parch            Ticket      Fare  \\\n",
      "0           0       3    male  22.0      1      0         A/5 21171    7.2500   \n",
      "1           1       1  female  38.0      1      0          PC 17599   71.2833   \n",
      "2           1       3  female  26.0      0      0  STON/O2. 3101282    7.9250   \n",
      "3           1       1  female  35.0      1      0            113803   53.1000   \n",
      "4           0       3    male  35.0      0      0            373450    8.0500   \n",
      "5           0       3    male   NaN      0      0            330877    8.4583   \n",
      "6           0       1    male  54.0      0      0             17463   51.8625   \n",
      "7           0       3    male   2.0      3      1            349909   21.0750   \n",
      "8           1       3  female  27.0      0      2            347742   11.1333   \n",
      "9           1       2  female  14.0      1      0            237736   30.0708   \n",
      "10          1       3  female   4.0      1      1           PP 9549   16.7000   \n",
      "11          1       1  female  58.0      0      0            113783   26.5500   \n",
      "12          0       3    male  20.0      0      0         A/5. 2151    8.0500   \n",
      "13          0       3    male  39.0      1      5            347082   31.2750   \n",
      "14          0       3  female  14.0      0      0            350406    7.8542   \n",
      "15          1       2  female  55.0      0      0            248706   16.0000   \n",
      "16          0       3    male   2.0      4      1            382652   29.1250   \n",
      "17          1       2    male   NaN      0      0            244373   13.0000   \n",
      "18          0       3  female  31.0      1      0            345763   18.0000   \n",
      "19          1       3  female   NaN      0      0              2649    7.2250   \n",
      "20          0       2    male  35.0      0      0            239865   26.0000   \n",
      "21          1       2    male  34.0      0      0            248698   13.0000   \n",
      "22          1       3  female  15.0      0      0            330923    8.0292   \n",
      "23          1       1    male  28.0      0      0            113788   35.5000   \n",
      "24          0       3  female   8.0      3      1            349909   21.0750   \n",
      "25          1       3  female  38.0      1      5            347077   31.3875   \n",
      "26          0       3    male   NaN      0      0              2631    7.2250   \n",
      "27          0       1    male  19.0      3      2             19950  263.0000   \n",
      "28          1       3  female   NaN      0      0            330959    7.8792   \n",
      "29          0       3    male   NaN      0      0            349216    7.8958   \n",
      "..        ...     ...     ...   ...    ...    ...               ...       ...   \n",
      "861         0       2    male  21.0      1      0             28134   11.5000   \n",
      "862         1       1  female  48.0      0      0             17466   25.9292   \n",
      "863         0       3  female   NaN      8      2          CA. 2343   69.5500   \n",
      "864         0       2    male  24.0      0      0            233866   13.0000   \n",
      "865         1       2  female  42.0      0      0            236852   13.0000   \n",
      "866         1       2  female  27.0      1      0     SC/PARIS 2149   13.8583   \n",
      "867         0       1    male  31.0      0      0          PC 17590   50.4958   \n",
      "868         0       3    male   NaN      0      0            345777    9.5000   \n",
      "869         1       3    male   4.0      1      1            347742   11.1333   \n",
      "870         0       3    male  26.0      0      0            349248    7.8958   \n",
      "871         1       1  female  47.0      1      1             11751   52.5542   \n",
      "872         0       1    male  33.0      0      0               695    5.0000   \n",
      "873         0       3    male  47.0      0      0            345765    9.0000   \n",
      "874         1       2  female  28.0      1      0         P/PP 3381   24.0000   \n",
      "875         1       3  female  15.0      0      0              2667    7.2250   \n",
      "876         0       3    male  20.0      0      0              7534    9.8458   \n",
      "877         0       3    male  19.0      0      0            349212    7.8958   \n",
      "878         0       3    male   NaN      0      0            349217    7.8958   \n",
      "879         1       1  female  56.0      0      1             11767   83.1583   \n",
      "880         1       2  female  25.0      0      1            230433   26.0000   \n",
      "881         0       3    male  33.0      0      0            349257    7.8958   \n",
      "882         0       3  female  22.0      0      0              7552   10.5167   \n",
      "883         0       2    male  28.0      0      0  C.A./SOTON 34068   10.5000   \n",
      "884         0       3    male  25.0      0      0   SOTON/OQ 392076    7.0500   \n",
      "885         0       3  female  39.0      0      5            382652   29.1250   \n",
      "886         0       2    male  27.0      0      0            211536   13.0000   \n",
      "887         1       1  female  19.0      0      0            112053   30.0000   \n",
      "888         0       3  female   NaN      1      2        W./C. 6607   23.4500   \n",
      "889         1       1    male  26.0      0      0            111369   30.0000   \n",
      "890         0       3    male  32.0      0      0            370376    7.7500   \n",
      "\n",
      "           Cabin Embarked  \n",
      "0            NaN        S  \n",
      "1            C85        C  \n",
      "2            NaN        S  \n",
      "3           C123        S  \n",
      "4            NaN        S  \n",
      "5            NaN        Q  \n",
      "6            E46        S  \n",
      "7            NaN        S  \n",
      "8            NaN        S  \n",
      "9            NaN        C  \n",
      "10            G6        S  \n",
      "11          C103        S  \n",
      "12           NaN        S  \n",
      "13           NaN        S  \n",
      "14           NaN        S  \n",
      "15           NaN        S  \n",
      "16           NaN        Q  \n",
      "17           NaN        S  \n",
      "18           NaN        S  \n",
      "19           NaN        C  \n",
      "20           NaN        S  \n",
      "21           D56        S  \n",
      "22           NaN        Q  \n",
      "23            A6        S  \n",
      "24           NaN        S  \n",
      "25           NaN        S  \n",
      "26           NaN        C  \n",
      "27   C23 C25 C27        S  \n",
      "28           NaN        Q  \n",
      "29           NaN        S  \n",
      "..           ...      ...  \n",
      "861          NaN        S  \n",
      "862          D17        S  \n",
      "863          NaN        S  \n",
      "864          NaN        S  \n",
      "865          NaN        S  \n",
      "866          NaN        C  \n",
      "867          A24        S  \n",
      "868          NaN        S  \n",
      "869          NaN        S  \n",
      "870          NaN        S  \n",
      "871          D35        S  \n",
      "872  B51 B53 B55        S  \n",
      "873          NaN        S  \n",
      "874          NaN        C  \n",
      "875          NaN        C  \n",
      "876          NaN        S  \n",
      "877          NaN        S  \n",
      "878          NaN        S  \n",
      "879          C50        C  \n",
      "880          NaN        S  \n",
      "881          NaN        S  \n",
      "882          NaN        S  \n",
      "883          NaN        S  \n",
      "884          NaN        S  \n",
      "885          NaN        Q  \n",
      "886          NaN        S  \n",
      "887          B42        S  \n",
      "888          NaN        S  \n",
      "889         C148        C  \n",
      "890          NaN        Q  \n",
      "\n",
      "[891 rows x 10 columns]\n"
     ]
    }
   ],
   "source": [
    "# drop name and ID (These features are unique for every person)\n",
    "drop_col = ['Name', 'PassengerId']\n",
    "train_set = whole_train_set.drop(drop_col, axis = 1)\n",
    "test_set = whole_test_set.drop(drop_col, axis = 1)\n",
    "print(train_set)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Age features is null about 25%, so this feature should be drop or replace with some values.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 210,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-0.0772210945722\n"
     ]
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "# 1 plot age drop null\n",
    "age_plot = train_set.loc[np.logical_not(train_set['Age'].isnull()), ['Age']]\n",
    "#sns.distplot(age_plot, bins = 50, axlabel = 'Age drop null', color=\"r\")\n",
    "\n",
    "# 2 correlation of age vs survival\n",
    "age_sur = train_set.loc[np.logical_not(train_set['Age'].isnull()) ,'Age'].corr(train_set.loc[np.logical_not(train_set['Age'].isnull()) ,'Survived'])\n",
    "print(age_sur)\n",
    "\n",
    "# 3 drop age\n",
    "train_set = train_set.drop(['Age'], axis = 1)\n",
    "test_set = test_set.drop(['Age'], axis = 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 211,
   "metadata": {
    "collapsed": true,
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# Find distinct values for categorical features\n",
    "choose_elements = ['Embarked', 'Ticket', 'Cabin', 'Parch']\n",
    "train_set_copy = train_set.copy()\n",
    "train_set_copy = train_set_copy[choose_elements]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 212,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Unique Value for featured Cabin : ['N' 'C' 'E' 'G' 'D' 'A' 'B' 'F' 'T']\n",
      "     Pclass     Sex  SibSp  Parch              Ticket      Fare Embarked  \\\n",
      "0         3    male      0      0              330911    7.8292        Q   \n",
      "1         3  female      1      0              363272    7.0000        S   \n",
      "2         2    male      0      0              240276    9.6875        Q   \n",
      "3         3    male      0      0              315154    8.6625        S   \n",
      "4         3  female      1      1             3101298   12.2875        S   \n",
      "5         3    male      0      0                7538    9.2250        S   \n",
      "6         3  female      0      0              330972    7.6292        Q   \n",
      "7         2    male      1      1              248738   29.0000        S   \n",
      "8         3  female      0      0                2657    7.2292        C   \n",
      "9         3    male      2      0           A/4 48871   24.1500        S   \n",
      "10        3    male      0      0              349220    7.8958        S   \n",
      "11        1    male      0      0                 694   26.0000        S   \n",
      "12        1  female      1      0               21228   82.2667        S   \n",
      "13        2    male      1      0               24065   26.0000        S   \n",
      "14        1  female      1      0         W.E.P. 5734   61.1750        S   \n",
      "15        2  female      1      0       SC/PARIS 2167   27.7208        C   \n",
      "16        2    male      0      0              233734   12.3500        Q   \n",
      "17        3    male      0      0                2692    7.2250        C   \n",
      "18        3  female      1      0    STON/O2. 3101270    7.9250        S   \n",
      "19        3  female      0      0                2696    7.2250        C   \n",
      "20        1    male      1      0            PC 17603   59.4000        C   \n",
      "21        3    male      0      1             C 17368    3.1708        S   \n",
      "22        1  female      0      0            PC 17598   31.6833        S   \n",
      "23        1    male      0      1            PC 17597   61.3792        C   \n",
      "24        1  female      1      3            PC 17608  262.3750        C   \n",
      "25        3    male      1      0           A/5. 3337   14.5000        S   \n",
      "26        1  female      0      1              113509   61.9792        C   \n",
      "27        3    male      0      0                2698    7.2250        C   \n",
      "28        1    male      0      0              113054   30.5000        S   \n",
      "29        3    male      2      0                2662   21.6792        C   \n",
      "..      ...     ...    ...    ...                 ...       ...      ...   \n",
      "388       3    male      0      0              364858    7.7500        Q   \n",
      "389       3    male      3      1              349909   21.0750        S   \n",
      "390       1    male      0      0               12749   93.5000        S   \n",
      "391       1  female      0      1            PC 17592   39.4000        S   \n",
      "392       3    male      0      2           C.A. 2673   20.2500        S   \n",
      "393       2    male      0      0          C.A. 30769   10.5000        S   \n",
      "394       3    male      3      1              315153   22.0250        S   \n",
      "395       1  female      1      0               13695   60.0000        S   \n",
      "396       3    male      0      0              371109    7.2500        Q   \n",
      "397       1  female      1      1               13567   79.2000        C   \n",
      "398       3    male      0      0              347065    7.7750        S   \n",
      "399       3    male      0      0               21332    7.7333        Q   \n",
      "400       1  female      0      0               36928  164.8667        S   \n",
      "401       2    male      1      0               28664   21.0000        S   \n",
      "402       1  female      0      1              112378   59.4000        C   \n",
      "403       1    male      0      0              113059   47.1000        S   \n",
      "404       1    male      1      0               17765   27.7208        C   \n",
      "405       2    male      0      0       SC/PARIS 2166   13.8625        C   \n",
      "406       2    male      1      0               28666   10.5000        S   \n",
      "407       1    male      1      1              113503  211.5000        C   \n",
      "408       3  female      0      0              334915    7.7208        Q   \n",
      "409       3  female      1      1  SOTON/O.Q. 3101315   13.7750        S   \n",
      "410       3  female      0      0              365237    7.7500        Q   \n",
      "411       1  female      1      0               19928   90.0000        Q   \n",
      "412       3  female      0      0              347086    7.7750        S   \n",
      "413       3    male      0      0           A.5. 3236    8.0500        S   \n",
      "414       1  female      0      0            PC 17758  108.9000        C   \n",
      "415       3    male      0      0  SOTON/O.Q. 3101262    7.2500        S   \n",
      "416       3    male      0      0              359309    8.0500        S   \n",
      "417       3    male      1      1                2668   22.3583        C   \n",
      "\n",
      "     Cabin_Numeric  \n",
      "0                0  \n",
      "1                0  \n",
      "2                0  \n",
      "3                0  \n",
      "4                0  \n",
      "5                0  \n",
      "6                0  \n",
      "7                0  \n",
      "8                0  \n",
      "9                0  \n",
      "10               0  \n",
      "11               0  \n",
      "12               6  \n",
      "13               0  \n",
      "14               2  \n",
      "15               0  \n",
      "16               0  \n",
      "17               0  \n",
      "18               0  \n",
      "19               0  \n",
      "20               0  \n",
      "21               0  \n",
      "22               0  \n",
      "23               0  \n",
      "24               6  \n",
      "25               0  \n",
      "26               6  \n",
      "27               0  \n",
      "28               5  \n",
      "29               0  \n",
      "..             ...  \n",
      "388              0  \n",
      "389              0  \n",
      "390              6  \n",
      "391              4  \n",
      "392              0  \n",
      "393              0  \n",
      "394              0  \n",
      "395              1  \n",
      "396              0  \n",
      "397              6  \n",
      "398              0  \n",
      "399              0  \n",
      "400              1  \n",
      "401              0  \n",
      "402              0  \n",
      "403              0  \n",
      "404              4  \n",
      "405              4  \n",
      "406              0  \n",
      "407              1  \n",
      "408              0  \n",
      "409              0  \n",
      "410              0  \n",
      "411              1  \n",
      "412              0  \n",
      "413              0  \n",
      "414              1  \n",
      "415              0  \n",
      "416              0  \n",
      "417              0  \n",
      "\n",
      "[418 rows x 8 columns]\n"
     ]
    }
   ],
   "source": [
    "## Deal with Cabin features -> change string feature to numeric feature\n",
    "# step 1 change null in column to N (no cabin)\n",
    "train_set.loc[train_set['Cabin'].isnull(), 'Cabin'] = 'N'\n",
    "test_set.loc[test_set['Cabin'].isnull(), 'Cabin'] = 'N'\n",
    "\n",
    "# step 2 change cabin type to abbreviate type (neglect number of cabin)\n",
    "train_set.loc[:, 'Cabin'] = train_set.loc[:, 'Cabin'].map(lambda x: str(x)[0])\n",
    "test_set.loc[:, 'Cabin'] = test_set.loc[:, 'Cabin'].map(lambda x: str(x)[0])\n",
    "\n",
    "# step 3 identify unique value for cabin features\n",
    "print('Unique Value for featured Cabin :',train_set['Cabin'].unique())\n",
    "\n",
    "# step 4 replace categorical with features\n",
    "dic_cabin = {'N':0, 'C':1, 'E':2, 'G':3, 'D':4, 'A':5, 'B':6, 'F':7, 'T':8}\n",
    "train_set['Cabin_Numeric'] = train_set['Cabin'].map(dic_cabin)\n",
    "test_set['Cabin_Numeric'] = test_set['Cabin'].map(dic_cabin)\n",
    "\n",
    "# step 5 drop cabin\n",
    "train_set = train_set.drop(['Cabin'], axis = 1)\n",
    "test_set = test_set.drop(['Cabin'], axis = 1)\n",
    "\n",
    "print(test_set)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 213,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "     Pclass     Sex  SibSp  Parch              Ticket      Fare  \\\n",
      "0         3    male      0      0              330911    7.8292   \n",
      "1         3  female      1      0              363272    7.0000   \n",
      "2         2    male      0      0              240276    9.6875   \n",
      "3         3    male      0      0              315154    8.6625   \n",
      "4         3  female      1      1             3101298   12.2875   \n",
      "5         3    male      0      0                7538    9.2250   \n",
      "6         3  female      0      0              330972    7.6292   \n",
      "7         2    male      1      1              248738   29.0000   \n",
      "8         3  female      0      0                2657    7.2292   \n",
      "9         3    male      2      0           A/4 48871   24.1500   \n",
      "10        3    male      0      0              349220    7.8958   \n",
      "11        1    male      0      0                 694   26.0000   \n",
      "12        1  female      1      0               21228   82.2667   \n",
      "13        2    male      1      0               24065   26.0000   \n",
      "14        1  female      1      0         W.E.P. 5734   61.1750   \n",
      "15        2  female      1      0       SC/PARIS 2167   27.7208   \n",
      "16        2    male      0      0              233734   12.3500   \n",
      "17        3    male      0      0                2692    7.2250   \n",
      "18        3  female      1      0    STON/O2. 3101270    7.9250   \n",
      "19        3  female      0      0                2696    7.2250   \n",
      "20        1    male      1      0            PC 17603   59.4000   \n",
      "21        3    male      0      1             C 17368    3.1708   \n",
      "22        1  female      0      0            PC 17598   31.6833   \n",
      "23        1    male      0      1            PC 17597   61.3792   \n",
      "24        1  female      1      3            PC 17608  262.3750   \n",
      "25        3    male      1      0           A/5. 3337   14.5000   \n",
      "26        1  female      0      1              113509   61.9792   \n",
      "27        3    male      0      0                2698    7.2250   \n",
      "28        1    male      0      0              113054   30.5000   \n",
      "29        3    male      2      0                2662   21.6792   \n",
      "..      ...     ...    ...    ...                 ...       ...   \n",
      "388       3    male      0      0              364858    7.7500   \n",
      "389       3    male      3      1              349909   21.0750   \n",
      "390       1    male      0      0               12749   93.5000   \n",
      "391       1  female      0      1            PC 17592   39.4000   \n",
      "392       3    male      0      2           C.A. 2673   20.2500   \n",
      "393       2    male      0      0          C.A. 30769   10.5000   \n",
      "394       3    male      3      1              315153   22.0250   \n",
      "395       1  female      1      0               13695   60.0000   \n",
      "396       3    male      0      0              371109    7.2500   \n",
      "397       1  female      1      1               13567   79.2000   \n",
      "398       3    male      0      0              347065    7.7750   \n",
      "399       3    male      0      0               21332    7.7333   \n",
      "400       1  female      0      0               36928  164.8667   \n",
      "401       2    male      1      0               28664   21.0000   \n",
      "402       1  female      0      1              112378   59.4000   \n",
      "403       1    male      0      0              113059   47.1000   \n",
      "404       1    male      1      0               17765   27.7208   \n",
      "405       2    male      0      0       SC/PARIS 2166   13.8625   \n",
      "406       2    male      1      0               28666   10.5000   \n",
      "407       1    male      1      1              113503  211.5000   \n",
      "408       3  female      0      0              334915    7.7208   \n",
      "409       3  female      1      1  SOTON/O.Q. 3101315   13.7750   \n",
      "410       3  female      0      0              365237    7.7500   \n",
      "411       1  female      1      0               19928   90.0000   \n",
      "412       3  female      0      0              347086    7.7750   \n",
      "413       3    male      0      0           A.5. 3236    8.0500   \n",
      "414       1  female      0      0            PC 17758  108.9000   \n",
      "415       3    male      0      0  SOTON/O.Q. 3101262    7.2500   \n",
      "416       3    male      0      0              359309    8.0500   \n",
      "417       3    male      1      1                2668   22.3583   \n",
      "\n",
      "     Cabin_Numeric  Embarked_Numeric  \n",
      "0                0                 2  \n",
      "1                0                 0  \n",
      "2                0                 2  \n",
      "3                0                 0  \n",
      "4                0                 0  \n",
      "5                0                 0  \n",
      "6                0                 2  \n",
      "7                0                 0  \n",
      "8                0                 1  \n",
      "9                0                 0  \n",
      "10               0                 0  \n",
      "11               0                 0  \n",
      "12               6                 0  \n",
      "13               0                 0  \n",
      "14               2                 0  \n",
      "15               0                 1  \n",
      "16               0                 2  \n",
      "17               0                 1  \n",
      "18               0                 0  \n",
      "19               0                 1  \n",
      "20               0                 1  \n",
      "21               0                 0  \n",
      "22               0                 0  \n",
      "23               0                 1  \n",
      "24               6                 1  \n",
      "25               0                 0  \n",
      "26               6                 1  \n",
      "27               0                 1  \n",
      "28               5                 0  \n",
      "29               0                 1  \n",
      "..             ...               ...  \n",
      "388              0                 2  \n",
      "389              0                 0  \n",
      "390              6                 0  \n",
      "391              4                 0  \n",
      "392              0                 0  \n",
      "393              0                 0  \n",
      "394              0                 0  \n",
      "395              1                 0  \n",
      "396              0                 2  \n",
      "397              6                 1  \n",
      "398              0                 0  \n",
      "399              0                 2  \n",
      "400              1                 0  \n",
      "401              0                 0  \n",
      "402              0                 1  \n",
      "403              0                 0  \n",
      "404              4                 1  \n",
      "405              4                 1  \n",
      "406              0                 0  \n",
      "407              1                 1  \n",
      "408              0                 2  \n",
      "409              0                 0  \n",
      "410              0                 2  \n",
      "411              1                 2  \n",
      "412              0                 0  \n",
      "413              0                 0  \n",
      "414              1                 1  \n",
      "415              0                 0  \n",
      "416              0                 0  \n",
      "417              0                 1  \n",
      "\n",
      "[418 rows x 8 columns]\n"
     ]
    }
   ],
   "source": [
    "## Deal with embarked features -> change embark feature to numerical\n",
    "# step 1 convert null value to unknown\n",
    "train_set.loc[train_set['Embarked'].isnull(),'Embarked'] = 'unknown'\n",
    "test_set.loc[test_set['Embarked'].isnull(),'Embarked'] = 'unknown'\n",
    "\n",
    "# step 2 convert string to numeric value\n",
    "dic_embarked = {'S':0, 'C':1, 'Q':2, 'unknown':3}\n",
    "train_set['Embarked_Numeric'] = train_set['Embarked'].map(dic_embarked)\n",
    "test_set['Embarked_Numeric'] = test_set['Embarked'].map(dic_embarked)\n",
    "\n",
    "# step 3 drop embarked feature\n",
    "train_set = train_set.drop(['Embarked'], axis = 1)\n",
    "test_set = test_set.drop(['Embarked'], axis = 1)\n",
    "\n",
    "print(test_set)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 214,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "     Survived  Pclass  SibSp  Parch            Ticket      Fare  \\\n",
      "0           0       3      1      0         A/5 21171    7.2500   \n",
      "1           1       1      1      0          PC 17599   71.2833   \n",
      "2           1       3      0      0  STON/O2. 3101282    7.9250   \n",
      "3           1       1      1      0            113803   53.1000   \n",
      "4           0       3      0      0            373450    8.0500   \n",
      "5           0       3      0      0            330877    8.4583   \n",
      "6           0       1      0      0             17463   51.8625   \n",
      "7           0       3      3      1            349909   21.0750   \n",
      "8           1       3      0      2            347742   11.1333   \n",
      "9           1       2      1      0            237736   30.0708   \n",
      "10          1       3      1      1           PP 9549   16.7000   \n",
      "11          1       1      0      0            113783   26.5500   \n",
      "12          0       3      0      0         A/5. 2151    8.0500   \n",
      "13          0       3      1      5            347082   31.2750   \n",
      "14          0       3      0      0            350406    7.8542   \n",
      "15          1       2      0      0            248706   16.0000   \n",
      "16          0       3      4      1            382652   29.1250   \n",
      "17          1       2      0      0            244373   13.0000   \n",
      "18          0       3      1      0            345763   18.0000   \n",
      "19          1       3      0      0              2649    7.2250   \n",
      "20          0       2      0      0            239865   26.0000   \n",
      "21          1       2      0      0            248698   13.0000   \n",
      "22          1       3      0      0            330923    8.0292   \n",
      "23          1       1      0      0            113788   35.5000   \n",
      "24          0       3      3      1            349909   21.0750   \n",
      "25          1       3      1      5            347077   31.3875   \n",
      "26          0       3      0      0              2631    7.2250   \n",
      "27          0       1      3      2             19950  263.0000   \n",
      "28          1       3      0      0            330959    7.8792   \n",
      "29          0       3      0      0            349216    7.8958   \n",
      "..        ...     ...    ...    ...               ...       ...   \n",
      "861         0       2      1      0             28134   11.5000   \n",
      "862         1       1      0      0             17466   25.9292   \n",
      "863         0       3      8      2          CA. 2343   69.5500   \n",
      "864         0       2      0      0            233866   13.0000   \n",
      "865         1       2      0      0            236852   13.0000   \n",
      "866         1       2      1      0     SC/PARIS 2149   13.8583   \n",
      "867         0       1      0      0          PC 17590   50.4958   \n",
      "868         0       3      0      0            345777    9.5000   \n",
      "869         1       3      1      1            347742   11.1333   \n",
      "870         0       3      0      0            349248    7.8958   \n",
      "871         1       1      1      1             11751   52.5542   \n",
      "872         0       1      0      0               695    5.0000   \n",
      "873         0       3      0      0            345765    9.0000   \n",
      "874         1       2      1      0         P/PP 3381   24.0000   \n",
      "875         1       3      0      0              2667    7.2250   \n",
      "876         0       3      0      0              7534    9.8458   \n",
      "877         0       3      0      0            349212    7.8958   \n",
      "878         0       3      0      0            349217    7.8958   \n",
      "879         1       1      0      1             11767   83.1583   \n",
      "880         1       2      0      1            230433   26.0000   \n",
      "881         0       3      0      0            349257    7.8958   \n",
      "882         0       3      0      0              7552   10.5167   \n",
      "883         0       2      0      0  C.A./SOTON 34068   10.5000   \n",
      "884         0       3      0      0   SOTON/OQ 392076    7.0500   \n",
      "885         0       3      0      5            382652   29.1250   \n",
      "886         0       2      0      0            211536   13.0000   \n",
      "887         1       1      0      0            112053   30.0000   \n",
      "888         0       3      1      2        W./C. 6607   23.4500   \n",
      "889         1       1      0      0            111369   30.0000   \n",
      "890         0       3      0      0            370376    7.7500   \n",
      "\n",
      "     Cabin_Numeric  Embarked_Numeric  Sex_Numeric  \n",
      "0                0                 0            0  \n",
      "1                1                 1            1  \n",
      "2                0                 0            1  \n",
      "3                1                 0            1  \n",
      "4                0                 0            0  \n",
      "5                0                 2            0  \n",
      "6                2                 0            0  \n",
      "7                0                 0            0  \n",
      "8                0                 0            1  \n",
      "9                0                 1            1  \n",
      "10               3                 0            1  \n",
      "11               1                 0            1  \n",
      "12               0                 0            0  \n",
      "13               0                 0            0  \n",
      "14               0                 0            1  \n",
      "15               0                 0            1  \n",
      "16               0                 2            0  \n",
      "17               0                 0            0  \n",
      "18               0                 0            1  \n",
      "19               0                 1            1  \n",
      "20               0                 0            0  \n",
      "21               4                 0            0  \n",
      "22               0                 2            1  \n",
      "23               5                 0            0  \n",
      "24               0                 0            1  \n",
      "25               0                 0            1  \n",
      "26               0                 1            0  \n",
      "27               1                 0            0  \n",
      "28               0                 2            1  \n",
      "29               0                 0            0  \n",
      "..             ...               ...          ...  \n",
      "861              0                 0            0  \n",
      "862              4                 0            1  \n",
      "863              0                 0            1  \n",
      "864              0                 0            0  \n",
      "865              0                 0            1  \n",
      "866              0                 1            1  \n",
      "867              5                 0            0  \n",
      "868              0                 0            0  \n",
      "869              0                 0            0  \n",
      "870              0                 0            0  \n",
      "871              4                 0            1  \n",
      "872              6                 0            0  \n",
      "873              0                 0            0  \n",
      "874              0                 1            1  \n",
      "875              0                 1            1  \n",
      "876              0                 0            0  \n",
      "877              0                 0            0  \n",
      "878              0                 0            0  \n",
      "879              1                 1            1  \n",
      "880              0                 0            1  \n",
      "881              0                 0            0  \n",
      "882              0                 0            1  \n",
      "883              0                 0            0  \n",
      "884              0                 0            0  \n",
      "885              0                 2            1  \n",
      "886              0                 0            0  \n",
      "887              6                 0            1  \n",
      "888              0                 0            1  \n",
      "889              1                 1            0  \n",
      "890              0                 2            0  \n",
      "\n",
      "[891 rows x 9 columns]\n"
     ]
    }
   ],
   "source": [
    "## Deal with sex features\n",
    "# convert string to numeric value\n",
    "dic_sex = {'male':0, 'female':1}\n",
    "train_set['Sex_Numeric'] = train_set['Sex'].map(dic_sex)\n",
    "test_set['Sex_Numeric'] = train_set['Sex'].map(dic_sex)\n",
    "\n",
    "# drop sex features\n",
    "train_set = train_set.drop(['Sex'], axis = 1)\n",
    "test_set = test_set.drop(['Sex'], axis = 1)\n",
    "print(train_set)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 215,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "     Pclass  SibSp  Parch      Fare  Cabin_Numeric  Embarked_Numeric  \\\n",
      "0         3      0      0    7.8292              0                 2   \n",
      "1         3      1      0    7.0000              0                 0   \n",
      "2         2      0      0    9.6875              0                 2   \n",
      "3         3      0      0    8.6625              0                 0   \n",
      "4         3      1      1   12.2875              0                 0   \n",
      "5         3      0      0    9.2250              0                 0   \n",
      "6         3      0      0    7.6292              0                 2   \n",
      "7         2      1      1   29.0000              0                 0   \n",
      "8         3      0      0    7.2292              0                 1   \n",
      "9         3      2      0   24.1500              0                 0   \n",
      "10        3      0      0    7.8958              0                 0   \n",
      "11        1      0      0   26.0000              0                 0   \n",
      "12        1      1      0   82.2667              6                 0   \n",
      "13        2      1      0   26.0000              0                 0   \n",
      "14        1      1      0   61.1750              2                 0   \n",
      "15        2      1      0   27.7208              0                 1   \n",
      "16        2      0      0   12.3500              0                 2   \n",
      "17        3      0      0    7.2250              0                 1   \n",
      "18        3      1      0    7.9250              0                 0   \n",
      "19        3      0      0    7.2250              0                 1   \n",
      "20        1      1      0   59.4000              0                 1   \n",
      "21        3      0      1    3.1708              0                 0   \n",
      "22        1      0      0   31.6833              0                 0   \n",
      "23        1      0      1   61.3792              0                 1   \n",
      "24        1      1      3  262.3750              6                 1   \n",
      "25        3      1      0   14.5000              0                 0   \n",
      "26        1      0      1   61.9792              6                 1   \n",
      "27        3      0      0    7.2250              0                 1   \n",
      "28        1      0      0   30.5000              5                 0   \n",
      "29        3      2      0   21.6792              0                 1   \n",
      "..      ...    ...    ...       ...            ...               ...   \n",
      "388       3      0      0    7.7500              0                 2   \n",
      "389       3      3      1   21.0750              0                 0   \n",
      "390       1      0      0   93.5000              6                 0   \n",
      "391       1      0      1   39.4000              4                 0   \n",
      "392       3      0      2   20.2500              0                 0   \n",
      "393       2      0      0   10.5000              0                 0   \n",
      "394       3      3      1   22.0250              0                 0   \n",
      "395       1      1      0   60.0000              1                 0   \n",
      "396       3      0      0    7.2500              0                 2   \n",
      "397       1      1      1   79.2000              6                 1   \n",
      "398       3      0      0    7.7750              0                 0   \n",
      "399       3      0      0    7.7333              0                 2   \n",
      "400       1      0      0  164.8667              1                 0   \n",
      "401       2      1      0   21.0000              0                 0   \n",
      "402       1      0      1   59.4000              0                 1   \n",
      "403       1      0      0   47.1000              0                 0   \n",
      "404       1      1      0   27.7208              4                 1   \n",
      "405       2      0      0   13.8625              4                 1   \n",
      "406       2      1      0   10.5000              0                 0   \n",
      "407       1      1      1  211.5000              1                 1   \n",
      "408       3      0      0    7.7208              0                 2   \n",
      "409       3      1      1   13.7750              0                 0   \n",
      "410       3      0      0    7.7500              0                 2   \n",
      "411       1      1      0   90.0000              1                 2   \n",
      "412       3      0      0    7.7750              0                 0   \n",
      "413       3      0      0    8.0500              0                 0   \n",
      "414       1      0      0  108.9000              1                 1   \n",
      "415       3      0      0    7.2500              0                 0   \n",
      "416       3      0      0    8.0500              0                 0   \n",
      "417       3      1      1   22.3583              0                 1   \n",
      "\n",
      "     Sex_Numeric  \n",
      "0              0  \n",
      "1              1  \n",
      "2              1  \n",
      "3              1  \n",
      "4              0  \n",
      "5              0  \n",
      "6              0  \n",
      "7              0  \n",
      "8              1  \n",
      "9              1  \n",
      "10             1  \n",
      "11             1  \n",
      "12             0  \n",
      "13             0  \n",
      "14             1  \n",
      "15             1  \n",
      "16             0  \n",
      "17             0  \n",
      "18             1  \n",
      "19             1  \n",
      "20             0  \n",
      "21             0  \n",
      "22             1  \n",
      "23             0  \n",
      "24             1  \n",
      "25             1  \n",
      "26             0  \n",
      "27             0  \n",
      "28             1  \n",
      "29             0  \n",
      "..           ...  \n",
      "388            0  \n",
      "389            1  \n",
      "390            0  \n",
      "391            0  \n",
      "392            0  \n",
      "393            1  \n",
      "394            1  \n",
      "395            0  \n",
      "396            1  \n",
      "397            0  \n",
      "398            0  \n",
      "399            1  \n",
      "400            0  \n",
      "401            0  \n",
      "402            1  \n",
      "403            0  \n",
      "404            1  \n",
      "405            0  \n",
      "406            0  \n",
      "407            0  \n",
      "408            0  \n",
      "409            1  \n",
      "410            0  \n",
      "411            0  \n",
      "412            1  \n",
      "413            0  \n",
      "414            0  \n",
      "415            1  \n",
      "416            1  \n",
      "417            1  \n",
      "\n",
      "[418 rows x 7 columns]\n"
     ]
    }
   ],
   "source": [
    "## Deal with Ticket features\n",
    "# drop Ticket feature\n",
    "train_set = train_set.drop(['Ticket'], axis = 1)\n",
    "test_set = test_set.drop(['Ticket'], axis = 1)\n",
    "\n",
    "print(test_set)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 216,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{' Mr': 0, ' Mrs': 1, ' Miss': 2, ' Master': 3, ' Don': 4, ' Rev': 5, ' Dr': 6, ' Mme': 7, ' Ms': 8, ' Major': 9, ' Lady': 10, ' Sir': 11, ' Mlle': 12, ' Col': 13, ' Capt': 14, ' the Countess': 15, ' Jonkheer': 16}\n",
      "0      0\n",
      "1      1\n",
      "2      0\n",
      "3      0\n",
      "4      1\n",
      "5      0\n",
      "6      2\n",
      "7      0\n",
      "8      1\n",
      "9      0\n",
      "10     0\n",
      "11     0\n",
      "12     1\n",
      "13     0\n",
      "14     1\n",
      "15     1\n",
      "16     0\n",
      "17     0\n",
      "18     2\n",
      "19     1\n",
      "20     0\n",
      "21     3\n",
      "22     1\n",
      "23     0\n",
      "24     1\n",
      "25     0\n",
      "26     2\n",
      "27     0\n",
      "28     0\n",
      "29     0\n",
      "      ..\n",
      "388    0\n",
      "389    3\n",
      "390    0\n",
      "391    1\n",
      "392    3\n",
      "393    0\n",
      "394    0\n",
      "395    1\n",
      "396    0\n",
      "397    1\n",
      "398    0\n",
      "399    0\n",
      "400    2\n",
      "401    0\n",
      "402    2\n",
      "403    0\n",
      "404    0\n",
      "405    0\n",
      "406    0\n",
      "407    0\n",
      "408    2\n",
      "409    2\n",
      "410    2\n",
      "411    1\n",
      "412    2\n",
      "413    0\n",
      "414    0\n",
      "415    0\n",
      "416    0\n",
      "417    3\n",
      "Name: Name_Numeric, Length: 418, dtype: int64\n"
     ]
    }
   ],
   "source": [
    "## Deal with name features (extract Mr, Mrs or Sir from the name)\n",
    "\n",
    "# for train set\n",
    "name_split = whole_train_set['Name'].map(lambda x: x.split(','))\n",
    "name_split_temp = name_split.map(lambda x: x[1].split('.'))\n",
    "train_abb_name = name_split_temp.map(lambda x: x[0])\n",
    "train_abb_name_unique = train_abb_name.unique()\n",
    "\n",
    "\n",
    "# for test set\n",
    "name_split = whole_test_set['Name'].map(lambda x: x.split(','))\n",
    "name_split_temp = name_split.map(lambda x: x[1].split('.'))\n",
    "test_abb_name = name_split_temp.map(lambda x: x[0])\n",
    "test_abb_name_unique = test_abb_name.unique()\n",
    "\n",
    "# create convert dict\n",
    "dict_mr = {}\n",
    "for i, col in enumerate(train_abb_name_unique):\n",
    "    dict_mr[col] = int(i)\n",
    "print(dict_mr)\n",
    "\n",
    "train_abb_name_numeric = train_abb_name.map(dict_mr)\n",
    "test_abb_name_numeric = test_abb_name.map(dict_mr)\n",
    "\n",
    "train_set['Name_Numeric'] = train_abb_name_numeric\n",
    "test_abb_name_numeric[test_abb_name_numeric.isnull()] = 0 \n",
    "test_abb_name_numeric = test_abb_name_numeric.map(lambda x : int(x))\n",
    "test_set['Name_Numeric'] = test_abb_name_numeric\n",
    "\n",
    "print(test_set['Name_Numeric'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Visualize features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 217,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x2c4073ed668>"
      ]
     },
     "execution_count": 217,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAuIAAALACAYAAADWhxcCAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nG19Cihd7AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c4073ed748>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Visualize correlation of each features\n",
    "plt.figure(figsize=(12,12))\n",
    "plt.title('Pearson Correlation of Features')\n",
    "sns.heatmap(train_set.astype(float).corr(),linewidths=0.1,vmax=1.0, square=True, linecolor='white', annot=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 218,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Survived\n",
      "Survived Min : 0.00, Max : 1.00, Mean : 0.38, Median : 0.00, Std : 0.49\n",
      "Pclass\n",
      "Pclass Min : 1.00, Max : 3.00, Mean : 2.31, Median : 3.00, Std : 0.84\n",
      "SibSp\n",
      "SibSp Min : 0.00, Max : 8.00, Mean : 0.52, Median : 0.00, Std : 1.10\n",
      "Parch\n",
      "Parch Min : 0.00, Max : 6.00, Mean : 0.38, Median : 0.00, Std : 0.81\n",
      "Fare\n",
      "Fare Min : 0.00, Max : 512.33, Mean : 32.20, Median : 14.45, Std : 49.67\n",
      "Cabin_Numeric\n",
      "Cabin_Numeric Min : 0.00, Max : 8.00, Mean : 0.81, Median : 0.00, Std : 1.81\n",
      "Embarked_Numeric\n",
      "Embarked_Numeric Min : 0.00, Max : 3.00, Mean : 0.37, Median : 0.00, Std : 0.65\n",
      "Sex_Numeric\n",
      "Sex_Numeric Min : 0.00, Max : 1.00, Mean : 0.35, Median : 0.00, Std : 0.48\n",
      "Name_Numeric\n",
      "Name_Numeric Min : 0.00, Max : 16.00, Mean : 0.94, Median : 0.00, Std : 1.72\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXUAAAEFCAYAAAAc33cJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c406f46eb8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXUAAAEFCAYAAAAc33cJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c4075b8668>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXUAAAEFCAYAAAAc33cJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAFPdJREFUeJzt3XuQnXV9x/H3JguJoZuY6iJaqZdWv+IFHMFrDKQWiHFE\n1HqbjlZMG6GNAzoKcgnO6IBgRdSIjHYxBm8z1SBe0olgVRBSFRG0ROHLRGnrVJ2uGEI0JJhk+8fz\nLBzX3bPnZM/JOfvj/ZrJzHOe2/mcze7n/M7v3AbGxsaQJJVhTq8DSJI6x1KXpIJY6pJUEEtdkgpi\nqUtSQSx1SSrIYK8DaPaKiMcDPwVuq1fNAX4PfDgzP1Xv8x5g6/jlKc7zLuBHmfnlSbY9cHxEjAHD\nmfnrNjI+G/j7zDwtIo4Bzs7MV7V6/P6IiLnAF4EjgLWZeVmTfZ8HXAQ8gurn93PgHZn548a8EbEe\n2JKZl0xyjkcDHwKeCowB9wHvneznqfJZ6pqp+zLzmeMXIuJxwDci4neZeVVmvquFc7wI+MlkG1o8\nvpmnAY+tz3Uz0NVCr/0ZsBw4JDP3TrVTRMwDNgInZuYt9brXA5si4glt5L0C+PfMfG19jqcCmyPi\nBZl5+wxvi2YZS10dlZn/XY+8zwSuahxhRsS7gVcA9wN3A6cArwSOAd4fEXuBk4E/Bf6CqvAexR+O\nUC+sR99zgDWZuTEiTgFelZkvBRi/DPwj8B5gUUR8ErgSuCwznx4Ri4CPAs+kGt1uAs7NzD0RsQu4\nGDgBeAzVI48PTbytEbEUeD+woL5Na4DNwNeAg4AfRMTfZOZPG445BriiviNcADwc+JOG034WuBeY\nW5//ssx8er3thRHxKmAhcC3ViH4P8GjgYRExJzP3ZeZPIuJlwLb6OvdQjeT/Cjikvp1fnOr/ULOb\nc+rqhh8Bz2hcERGHA28Fnp2Zx1CV0nMz86PAzcCZmXl1vfuCzHxaZr5zknP/LDOfBbweuDIihqcK\nkZk/B94F3JCZb5qweS3VHcszqO5UjgLeUW+bB/w6M5dQ3TlcHBHzJ9yeRwAbgDMy80jgjcBngEcC\nL6F+BNNY6HWmm8cf2WTmNuAs4GsR8bOI+DTwJqpR9/2T3KTHAn9NdUd0FLCqXv8O4C3A/0XElyPi\nzPrn9Kt6+1zgN5l5NPAaYF2zn5tmN0td3TAG7Jyw7n+pyv6WiLgE+GFmfmmK429scu6PAWTmFqop\nm+fvZ8YVVKPgsczcXZ93RcP28fnoW6hK/pAJxz+Xaq7/e3WeH1ON0pe1EyIzL6V6NHI68EvgncCt\n9SOJiT6dmb+rC/8zVI8kyMxvAn8OvBz4HnAScEf9iGbcZfW+/0n1HMix7eTU7GGpqxuezYNPngKQ\nmfuA46imXO4GPhgRH57i+N82OXfjHPUA1ROzY/XyuINbyDjxd38O1ZTJuPsAMnP8w5EGJtl/snMe\nNMn6SUXEkog4MzN3ZObGzDyL6jmAfdSFPcEf3faIODQiLgfGMvPGzHxvZh4L/CvVo4dxeybknHKu\nX7Obpa6OiognA+cDH5iw/ihgC3B7Zl4EfJBqCgGqwmm1DE+pz/cs4ElUI9NR4OkRMT8iBqlGquOm\nOvc1wOqIGKifsHwz8PUWMwB8t4oRz6nzPI1q9HtdG+cYBdZExAsb1j2a6lHBbZPs/7qImFdPBZ1C\n9TzAb6juAM6IiIE6ywKqkfstDcf+Xb3tWcBTgOvbyKlZxFLXTD0sIn5Y/7sFWA+ck5n/1rhTZv4I\n+Dxwc0TcDKwE3lZv/ipwSUQ0jiyn8sSIuJXqFR+vy8zfUM3PXw/cAdzAHxbid4CnRMTVE85zOnBo\nve9tQAIXtnibqV9W+WrgIxFxG/A54E2ZeWez4yLimIj4YX2OO6mmTN5bz6n/hOpn9ObMzEkOv4tq\naupW4NvAlfUTpSdSTUPdFRFbqO7orsnMdQ3HLqn/f9YBr63n81WgAT96Vyrb/ry+X7OXI3VJKogj\ndUkqiCN1SSqIpS5JBenpxwSMju7Y77mfxYsXsG3bxPe39J652mOu9vVrNnO1Zya5hoeHJr5v4gGz\ndqQ+ODi31xEmZa72mKt9/ZrNXO3pVq5ZW+qSpD9mqUtSQSx1SSqIpS5JBbHUJakglrokFcRSl6SC\nWOqSVBBLXZIK0tLHBETEOcDLqL4m7HKqLyRYT/U1YluA1Zm5LyJWAadSfdvMBZm5sRuhAU56+5en\n36lL1p39op5dtyQ1M+1IPSKWAS8AllB9x+ThwKXAmsxcSvVdiSdHxGFU3yazBFgOXFR/TZgk6QBp\nZaS+nOrrvq4GFgJnAqt48DsON1F9ndZeYHP9zey7I2IrcCTw/alOvHjxgr79XIZmhoeHZrS9V8zV\nnn7NBf2bzVzt6UauVkr9kcDjgJcCTwC+Asxp+Jb1HcAiqsLf3nDc+Pop9eMnp7VidHTHlNuGh4ea\nbu8Vc7WnX3NB/2YzV3tmkqvZnUErpX43cEdm3g9kROyimoIZNwTcA9xbL09cL0k6QFp59cuNwIsj\nYiAiHgMcAnyjnmsHWEH1De43AUsjYn5ELAKOoHoSVZJ0gEw7Us/MjRFxLFVpzwFWA3cBIxFxMHA7\nsCEz90bEWqqCnwOcl5m7uhddkjRRSy9pzMyzJll93CT7jQAjMw0lSdo/vvlIkgpiqUtSQSx1SSqI\npS5JBbHUJakglrokFcRSl6SCWOqSVBBLXZIKYqlLUkEsdUkqiKUuSQWx1CWpIJa6JBXEUpekgljq\nklQQS12SCmKpS1JBLHVJKoilLkkFsdQlqSCWuiQVxFKXpIJY6pJUEEtdkgpiqUtSQQZb2SkibgHu\nrS/eBVwIrAfGgC3A6szcFxGrgFOBPcAFmbmx44klSVOattQjYj4wkJnLGtZ9BViTmddFxMeAkyPi\nO8DpwDHAfODGiPh6Zu7uTnRJ0kStjNSPAhZExLX1/ucCRwPX19s3AScCe4HNdYnvjoitwJHA96c6\n8eLFCxgcnDuD+L0xPDw0o+29Yq729Gsu6N9s5mpPN3K1Uuo7gUuAK4AnUZX4QGaO1dt3AIuAhcD2\nhuPG109p27ad7ebtC6OjO6bcNjw81HR7r5irPf2aC/o3m7naM5Ncze4MWin1O4GtdYnfGRF3U43U\nxw0B91DNuQ9Nsl6SdIC08uqXlcAHACLiMVQj8msjYlm9fQVwA3ATsDQi5kfEIuAIqidRJUkHSCsj\n9U8A6yPiRqpXu6wEfg2MRMTBwO3AhszcGxFrqQp+DnBeZu7qUm5J0iSmLfXMvB/420k2HTfJviPA\nSAdySZL2g28+kqSCWOqSVBBLXZIKYqlLUkEsdUkqiKUuSQWx1CWpIJa6JBXEUpekgljqklQQS12S\nCmKpS1JBLHVJKoilLkkFsdQlqSCWuiQVxFKXpIJY6pJUEEtdkgpiqUtSQSx1SSqIpS5JBbHUJakg\nlrokFcRSl6SCWOqSVJDBVnaKiEOBHwAnAHuA9cAYsAVYnZn7ImIVcGq9/YLM3NiVxJKkKU07Uo+I\ng4CPA/fVqy4F1mTmUmAAODkiDgNOB5YAy4GLImJedyJLkqbSyvTLJcDHgF/Ul48Grq+XNwHHA88B\nNmfm7szcDmwFjuxwVknSNJpOv0TEKcBoZl4TEefUqwcyc6xe3gEsAhYC2xsOHV/f1OLFCxgcnNt2\n6F4bHh6a0fZeMVd7+jUX9G82c7WnG7mmm1NfCYxFxPHAM4FPAYc2bB8C7gHurZcnrm9q27adbYXt\nF6OjO6bcNjw81HR7r5irPf2aC/o3m7naM5Ncze4MmpZ6Zh47vhwR1wGnAe+PiGWZeR2wAvgWcBNw\nYUTMB+YBR1A9iSpJOoBaevXLBG8HRiLiYOB2YENm7o2ItcANVPP052Xmrg7mlCS1oOVSz8xlDReP\nm2T7CDDSgUySpP3km48kqSCWuiQVxFKXpIJY6pJUEEtdkgpiqUtSQSx1SSqIpS5JBbHUJakglrok\nFcRSl6SCWOqSVBBLXZIKYqlLUkEsdUkqiKUuSQWx1CWpIJa6JBXEUpekgljqklQQS12SCmKpS1JB\nLHVJKoilLkkFsdQlqSCWuiQVZHC6HSJiLjACBDAGnAbsAtbXl7cAqzNzX0SsAk4F9gAXZObGLuWW\nJE2ilZH6SQCZuQRYA1wIXAqsycylwABwckQcBpwOLAGWAxdFxLyupJYkTWraUs/MLwFvri8+DrgH\nOBq4vl63CTgeeA6wOTN3Z+Z2YCtwZMcTS5KmNO30C0Bm7omIK4FXAK8CTsjMsXrzDmARsBDY3nDY\n+PopLV68gMHBuW2H7rXh4aEZbe8Vc7WnX3NB/2YzV3u6kaulUgfIzDdGxDuB7wEPa9g0RDV6v7de\nnrh+Stu27Ww9aR8ZHd0x5bbh4aGm23vFXO3p11zQv9nM1Z6Z5Gp2ZzDt9EtEvCEizqkv7gT2ATdH\nxLJ63QrgBuAmYGlEzI+IRcARVE+iSpIOkFZG6l8EPhkR3wYOAt4K3A6MRMTB9fKGzNwbEWupCn4O\ncF5m7upSbknSJKYt9cz8HfCaSTYdN8m+I1Qvf5Qk9YBvPpKkgljqklQQS12SCmKpS1JBLHVJKoil\nLkkFsdQlqSCWuiQVxFKXpIJY6pJUEEtdkgpiqUtSQSx1SSqIpS5JBbHUJakglrokFcRSl6SCWOqS\nVBBLXZIKYqlLUkEsdUkqiKUuSQWx1CWpIJa6JBXEUpekgljqklSQwWYbI+IgYB3weGAecAHwE2A9\nMAZsAVZn5r6IWAWcCuwBLsjMjd2L3VsrL/5mT6533dkv6sn1Spo9phupvx64OzOXAi8GLgMuBdbU\n6waAkyPiMOB0YAmwHLgoIuZ1L7YkaTJNR+rAF4AN9fIA1Sj8aOD6et0m4ERgL7A5M3cDuyNiK3Ak\n8P2OJ5YkTalpqWfmbwEiYoiq3NcAl2TmWL3LDmARsBDY3nDo+PqmFi9ewODg3P2I/dA0PDzU0+O7\nxVzt69ds5mpPN3JNN1InIg4HrgYuz8zPRcQ/N2weAu4B7q2XJ65vatu2ne2lfYgbHd2x38cODw/N\n6PhuMVf7+jWbudozk1zN7gyazqlHxKOAa4F3Zua6evWtEbGsXl4B3ADcBCyNiPkRsQg4gupJVEnS\nATTdSP1cYDFwfkScX687A1gbEQcDtwMbMnNvRKylKvg5wHmZuatboSVJk5tuTv0MqhKf6LhJ9h0B\nRjqUS5K0H3zzkSQVxFKXpIJY6pJUEEtdkgpiqUtSQSx1SSqIpS5JBbHUJakglrokFcRSl6SCWOqS\nVBBLXZIKYqlLUkEsdUkqiKUuSQWx1CWpIJa6JBXEUpekgljqklQQS12SCmKpS1JBLHVJKoilLkkF\nsdQlqSCWuiQVxFKXpIIMtrJTRDwXeF9mLouIvwTWA2PAFmB1Zu6LiFXAqcAe4ILM3NilzJKkKUw7\nUo+Is4ArgPn1qkuBNZm5FBgATo6Iw4DTgSXAcuCiiJjXnciSpKm0Mv3yU+CVDZePBq6vlzcBxwPP\nATZn5u7M3A5sBY7sZFBJ0vSmnX7JzKsi4vENqwYyc6xe3gEsAhYC2xv2GV/f1OLFCxgcnNt62oe4\n4eGhnh7fLeZqX79mM1d7upGrpTn1CfY1LA8B9wD31ssT1ze1bdvO/bj6h67R0R37fezw8NCMju8W\nc7WvX7OZqz0zydXszmB/Xv1ya0Qsq5dXADcANwFLI2J+RCwCjqB6ElWSdADtz0j97cBIRBwM3A5s\nyMy9EbGWquDnAOdl5q4O5pQktaClUs/M/wKeVy/fCRw3yT4jwEgnw0mS2uObjySpIJa6JBXEUpek\ngljqklQQS12SCmKpS1JBLHVJKoilLkkFsdQlqSCWuiQVxFKXpIJY6pJUEEtdkgpiqUtSQSx1SSqI\npS5JBbHUJakglrokFcRSl6SC7M8XT6tHVl78zZ5d97qzX9Sz65bUOkfqklQQS12SCmKpS1JBnFNX\nS3o1n9+rufyT3v7lnlwv+PyFZsaRuiQVxFKXpIJ0dPolIuYAlwNHAbuBf8jMrZ28DknqlF6+TPir\nHzi5K+ft9Jz6y4H5mfn8iHge8AGgO8mlQj3Unr9QZ3V6+uWFwNcAMvO7wDEdPr8kqYmBsbGxjp0s\nIq4ArsrMTfXl/wGemJl7OnYlkqQpdXqkfi8w1Hh+C12SDpxOl/pm4CUA9Zz6bR0+vySpiU4/UXo1\ncEJE/AcwALypw+eXJDXR0Tl1SVJv+eYjSSqIpS5JBbHUJakgs+5TGvv9owgi4rnA+zJzWa+zAETE\nQcA64PHAPOCCzPxKT0MBETEXGAECGANOy8wtvU31oIg4FPgBcEJm3tHrPAARcQvVy4YB7srMvngh\nQkScA7wMOBi4PDM/0eNIRMQpwCn1xfnAM4HDMvOeXmWCB/4er6T6e9wLrOr079dsHKk/8FEEwNlU\nH0XQFyLiLOAKql+ifvF64O7MXAq8GLisx3nGnQSQmUuANcCFvY3zoPoP7+PAfb3OMi4i5gMDmbms\n/tcvhb4MeAGwBDgOOLyngWqZuX78Z0V153x6rwu99hJgMDNfALyHLvzez8ZS7+ePIvgp8Mpeh5jg\nC8D59fIA0BdvBsvMLwFvri8+DuiHP7hxlwAfA37R6yANjgIWRMS1EfHN+n0g/WA51ftRrga+Cmzs\nbZw/FBHHAE/LzH/pdZbancBgPeOwEPh9p69gNpb6QmB7w+W9EdEX00iZeRVd+E+aicz8bWbuiIgh\nYAPVqLgvZOaeiLgS+Ajw2V7ngQceto9m5jW9zjLBTqo7m+XAacBn++T3/pFUA6tX82Cugd5G+gPn\nAu/udYgGv6WaermDavpxbaevYDaWuh9F0KaIOBz4FvDpzPxcr/M0ysw3Ak8GRiLikF7nAVZSvYHu\nOqp52E9FxGG9jQRUI7zPZOZYZt4J3A08useZoMpxTWben5kJ7AKGe5wJgIh4OBCZ+a1eZ2nwNqqf\n15OpHn1dWU+tdUw/3NO3azPVfOzn/SiC6UXEo4Brgbdk5jd6nWdcRLwBeGxmXkQ1Ct1X/+upzDx2\nfLku9tMy81e9S/SAlcAzgH+KiMdQPWL9ZW8jAXAjcEZEXEp1J3MIVdH3g2OBvvmdr23jwUfzvwEO\nAuZ28gpmY6n7UQTtORdYDJwfEeNz6ysys9dPAn4R+GREfJvqF/utfZCpn30CWB8RN1K9WmhlPzxC\nzcyNEXEscBPVI//Vmbm3x7HGBfCzXoeY4IPAuoi4gerVQudm5u86eQV+TIAkFWQ2zqlLkqZgqUtS\nQSx1SSqIpS5JBbHUJakglrokFcRSl6SC/D9Yr0Z9aUIHUQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c405286048>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXUAAAEFCAYAAAAc33cJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nFcRQl6SCGOqSVBBDXZIK8r8EaxYBtsew2AAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c404fff518>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXUAAAEFCAYAAAAc33cJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nwy/qiFuSAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c404dc3550>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXUAAAEFCAYAAAAc33cJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c4079718d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXUAAAEFCAYAAAAc33cJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nZHeZTcSrwgAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c4051647f0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXUAAAEFCAYAAAAc33cJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nVBFDXZIqYqhLUkUMdUmqiKEuSRX5XwOvuW7VPK4kAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c401853780>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXUAAAEFCAYAAAAc33cJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c404cff898>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "## visualize distribution of each features\n",
    "\n",
    "col_interest = ['Fare', 'SibSp', 'Parch']\n",
    "\n",
    "def r2(x): return round(x,2)\n",
    "\n",
    "for col in list(train_set):\n",
    "    fig = plt.figure()\n",
    "    ax1 = fig.gca()\n",
    "    \n",
    "    print(col)\n",
    "    plt.hist(train_set[col])\n",
    "    print('{:s} Min : {:.2f}, Max : {:.2f}, Mean : {:.2f}, Median : {:.2f}, Std : {:.2f}'.format\n",
    "          (col, np.min(train_set[col]), np.max(train_set[col]), np.mean(train_set[col])\n",
    "          , np.median(train_set[col]), np.std(train_set[col])))\n",
    "    plt.title('Distribution of :' + col)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 219,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c4073ed470>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAnQAAAKBCAYAAADeNjZlAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c404b2ecc0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "## visualize box plot\n",
    "\n",
    "col_interest = ['Fare']\n",
    "for col in col_interest:\n",
    "    fig = plt.figure()\n",
    "    train_set.boxplot(column = col, by = ['Survived'], figsize = (10,10))\n",
    "    plt.xlabel('Survived')\n",
    "    plt.ylabel(col)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 220,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.text.Text at 0x2c40541ac18>"
      ]
     },
     "execution_count": 220,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYMAAAEFCAYAAAABjYvXAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nMjiiAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c4073ed198>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYMAAAEFCAYAAAABjYvXAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nkjAMJEkYBpIkDANJEvD/Ad6JMwFDcw5oAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c4073ed7f0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "## Visualize histogram plot\n",
    "\n",
    "fig = plt.figure()\n",
    "train_set['Fare'].plot.hist(bins = 30)\n",
    "plt.title('Fare Distribuition of train set')\n",
    "\n",
    "fig = plt.figure()\n",
    "test_set['Fare'].plot.hist(bins = 30)\n",
    "plt.title('Fare Distribuition of test set')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 221,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dead\n",
      " 0    468\n",
      "1     81\n",
      "Name: Sex_Numeric, dtype: int64\n",
      "survived\n",
      " 0    109\n",
      "1    233\n",
      "Name: Sex_Numeric, dtype: int64\n",
      "0    549\n",
      "Name: Survived, dtype: int64\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmIAAAE4CAYAAAD4q5/qAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nJElFDGKSJElFDGKSJElF/g9Tj9g60mecdwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c406e43d68>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmIAAAE4CAYAAAD4q5/qAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAGptJREFUeJzt3X+U3XV95/FnyJDE1CFM10G6uxXqaX0femqkG4iUGAkC\nWjywAWs5u1YqsgW0uAHLrgIJS1tBsIX0NHAUO2kAXW1do3gwLb8UCJiuoPwqWekbYsvWLXKa1Uky\nmCYQMvvH95vuNCeTyUzmez/fufN8/JN7v/d77/f95s58eM3n+2vG8PAwkiRJ6rxDShcgSZI0XRnE\nJEmSCjGISZIkFWIQkyRJKsQgJkmSVIhBTJIkqZCe0gVoeoqIo4HvA0/Xiw4BXgH+ODM/N0nbeC/w\nkcxcMhmfJ2l6iIgTgOuAf0U1Nv0A+C+Z+b8m4bM/BByemddPwmcdB6zNzKMP9rNUjkFMJf1TZh67\n50lEHAV8MyJ+kplfKViXpGkqImYD64B3Zubj9bL3A3dFxM9l5qsH8/mZecsklKkuYhBTa2Tm/46I\n/wb814j4OvAp4CRgJvAEsCwzt0XEGcCVwCzgCOD2zLwKICJ+H/gN4EfAcwXakDS1zQUOB147YtkX\ngG3AKRGxMjN/CSAilgA3Z+YvRcTvAr8C/AywEVgMnJ2Z363X/XNgPfB64HXAncCNmfnm+vXDgb8D\n3ljXcDPwBuBQ4M8z85P1eh8GPgps5f/vUdAU5jFiapungDcDlwO7gAWZ+RbgBeD6iJgBXAZ8IDOP\nA04AroiI10XEUuDXgGOBE4F5JRqQNHVl5iDwMeDuiPjbiPg88EHgG8DLY7z9KODfZeb7gDXAeQAR\n0QecBnxxxLr3Aa+tdy8C/EfgL+rtfx5Yk5kLgIXAqRFxTkQcC/wu8PbMPP4A6tEUYBBT2wwD24Ez\ngKXAExHxJHAW8IuZOQycCSyIiKuBlcAM4KeAU4GvZuZQZu6iGgglaVwycyXVzNUy4IfAx6lm5cf6\n4+7b9dgD1fhzTkTMogpZX8/MrSO2MQz8KXVYowp7qyPip6j2BHyiHvu+TTUzdixwCnBvZr5Yv+dP\nDqZPtYO7JtU2x1NNt88DLsnMuwAi4rXAnHqQegK4A3iYarA7iyqMDdf/7rELSRqHiFgEnJiZf0h1\nrNi6iLiSalw6ln85xsza6+0v7XlQH2rxONUflR8ELt3H5m6l+mNzNdUB/A9GxGH1Nk7MzO11Ta8D\ndgAX4hjXdZwRU2tExJuAq4AbgXuAj0TErIg4BBigOovpF4DDgBWZ+XWqvxxnUx1Hdjfw6xFxeP2e\ncwu0IWlq2wysiIi3jVj2M1Sz7l8D3hARR9SHSZw1xmcNUM2mzc3MDXu/mJn/ADwCfBZYXS/bRjUL\n9jvwz8eObaDaQ3Af8M6I+Lf1R5w3kQbVLs6IqaTX1FPvALup/uK7IjP/IiLuB26gmv2aCTxJdWzY\nS1R/pf5NRGwBNgHfA34+M/8yIt4MfBcYpDrerL+TDUma2jLz2Yg4C/hkHXh2UB0Yf2FmPhURn6Ua\nY35INRbtz53Ap6lOPBrNALAW+Pcjlr0PuDkinqaadfuzzPwCQER8jOrs8iHg0XE3qNaZMTw8XLoG\nSZKkacldk5IkSYUYxCRJkgoxiEmSJBViEJMkSSpkSp41uXnzUONnGPT1zWVwcHvTm+kIe2mnbuml\nU3309/fOGHut9uvE+AXd8/MF3dNLt/QB9jJe+xu/nBEbRU/PzNIlTBp7aadu6aVb+ug23fS9dEsv\n3dIH2MtkMohJkiQVYhCTJEkqxCAmSZJUiEFMkiSpEIOYJElSIQYxSZKkQgxikiRJhTR6QdeIOAJ4\nDDgN2AXcBgwDG4GLM3N3RFwAXFS/fk1mrmuyJkmSpLZobEYsIg4FPgv8U71oJbAiMxcDM4ClEXEk\nsAxYBLwLuC4iZjdVkyRJUps0uWvyBuAW4IX6+QJgff34LuBUYCGwITN3ZuZWYBMwv8GaJEmSWqOR\nXZMRcR6wOTPviYgr6sUzMnPPPdaGgHnAYcDWEW/ds3y/+vrmduSWBP39vY1vo1PspZ26pZdu6aMT\nOjV+QXd9L93SS7f0AfYyWZo6Rux8YDgiTgWOBT4HHDHi9V5gC7Ctfrz38v2ayM05z7/+/nG/Z7zW\nXP6OxrcxEf39vWzePFS6jElhL+3TqT66ZdDv1I2Su+XnC7qnl27pA+xlItsYTSNBLDPfvudxRDwI\nfAj4w4hYkpkPAqcDDwCPAtdGxBxgNnAM1YH8kiRJXa/Rsyb3chkwEBGzgGeAtZn5akSsAh6mOl5t\neWbu6GBNkiRJxTQexDJzyYinJ+3j9QFgoOk6JEmS2sYLukqSJBViEJMkSSrEICZJklSIQUySJKkQ\ng5gkSVIhBjFJkqRCDGKSJEmFGMQkSZIKMYhJkiQVYhCTJEkqxCAmSZJUiEFMkiSpEIOYJElSIQYx\nSZKkQgxikiRJhRjEJEmSCjGISZIkFWIQkyRJKsQgJkmSVIhBTJIkqRCDmCRJUiEGMUmSpEIMYpIk\nSYUYxCRJkgoxiEmSJBXS09QHR8RMYAAIYBj4EHAosA54rl7tM5n5pYi4ALgI2AVck5nrmqpLkiSp\nLRoLYsCZAJm5KCKWANcCXwdWZuaNe1aKiCOBZcBxwBzgWxFxX2bubLA2SZKk4hoLYpn5tYjYM7N1\nFLAFWABERCylmhW7FFgIbKiD186I2ATMB77TVG2SJElt0OSMGJm5KyJuB84G3gv8G2B1Zj4WEcuB\nq4Enga0j3jYEzNvf5/b1zaWnZ2ZDVU9cf39v6RJG1ebaxste2qdb+uiETo5f3fS9dEsv3dIH2Mtk\naTSIAWTmByLi48AjwImZ+Q/1S3cANwEPASP/C/RSzZ6NanBwexOlHrTNm4dKl7BP/f29ra1tvOyl\nfTrVR7cM+p0av7rl5wu6p5du6QPsZSLbGE1jZ01GxLkRcUX9dDuwG/hqRCysl50CPAY8CiyOiDkR\nMQ84BtjYVF2SJElt0eSM2FeBWyPiIaqzJS8FfgDcFBGvAC8CF2bmtohYBTxMFQyXZ+aOBuuSJElq\nhSYP1v8JcM4+Xlq0j3UHqC51IUmSNG14QVdJkqRCDGKSJEmFGMQkSZIKMYhJkiQVYhCTJEkqxCAm\nSZJUiEFMkiSpEIOYJElSIQYxSZKkQgxikiRJhRjEJEmSCjGISZIkFWIQkyRJKsQgJkmSVIhBTJIk\nqRCDmCRJUiEGMUmSpEIMYpIkSYUYxCRJkgoxiEmSJBViEJMkSSrEICZJklSIQUySJKkQg5gkSVIh\nPU19cETMBAaAAIaBDwE7gNvq5xuBizNzd0RcAFwE7AKuycx1TdUlSZLUFk3OiJ0JkJmLgBXAtcBK\nYEVmLgZmAEsj4khgGbAIeBdwXUTMbrAuSZKkVmgsiGXm14AL66dHAVuABcD6etldwKnAQmBDZu7M\nzK3AJmB+U3VJkiS1RWO7JgEyc1dE3A6cDbwXOC0zh+uXh4B5wGHA1hFv27N8VH19c+npmdlAxQen\nv7+3dAmjanNt42Uv7dMtfXRCJ8evbvpeuqWXbukD7GWyNBrEADLzAxHxceAR4DUjXuqlmiXbVj/e\ne/moBge3T3aZk2Lz5qHSJexTf39va2sbL3tpn0710S2DfqfGr275+YLu6aVb+gB7mcg2RtPYrsmI\nODcirqifbgd2A9+NiCX1stOBh4FHgcURMSci5gHHUB3IL0mS1NWanBH7KnBrRDwEHApcCjwDDETE\nrPrx2sx8NSJWUYWyQ4DlmbmjwbokSZJaobEglpk/Ac7Zx0sn7WPdAapLXUiSJE0bXtBVkiSpEIOY\nJElSIQYxSZKkQgxikiRJhRjEJEmSCjGISZIkFWIQkyRJKsQgJkmSVIhBTJIkqRCDmCRJUiEGMUmS\npEIMYpIkSYUYxCRJkgoxiEmSJBViEJMkSSrEICZJklSIQUySJKkQg5gkSVIhBjFJkqRCDGKSJEmF\nGMQkSZIKMYhJkiQVYhCTJEkqxCAmSZJUiEFMkiSpkJ4mPjQiDgXWAEcDs4FrgB8A64Dn6tU+k5lf\niogLgIuAXcA1mbmuiZokSZLappEgBrwf+FFmnhsRPw08Cfw+sDIzb9yzUkQcCSwDjgPmAN+KiPsy\nc2dDdUmSJLVGU0Hsy8Da+vEMqtmuBUBExFKqWbFLgYXAhjp47YyITcB84DsN1SVJktQajQSxzHwJ\nICJ6qQLZCqpdlKsz87GIWA5cTTVTtnXEW4eAeWN9fl/fXHp6Zk563Qerv7+3dAmjanNt42Uv7dMt\nfXRCJ8evbvpeuqWXbukD7GWyNDUjRkT8LHAH8OnM/GJEHJ6ZW+qX7wBuAh4CRnbfC2xhDIOD2ye7\n3EmxefNQ6RL2qb+/t7W1jZe9tE+n+uiWQb9T41e3/HxB9/TSLX2AvUxkG6Np5KzJiHg9cC/w8cxc\nUy++JyIW1o9PAR4DHgUWR8SciJgHHANsbKImSZKktmlqRuxKoA+4KiKuqpf9DvBHEfEK8CJwYWZu\ni4hVwMNUoXB5Zu5oqCZJkqRWaeoYsUuAS/bx0qJ9rDsADDRRhyRJUpt5QVdJkqRCDGKSJEmFGMQk\nSZIKMYhJkiQVYhCTJEkqxCAmSZJUiEFMkiSpEIOYJElSIQYxSZKkQg74yvoR0QPMB3YBT2fmcGNV\nSZIkTQMHNCMWEacBfw/8CXA78LcRcXyThUmSJHW7A50R+yPg9Mx8CiAijgNuAY5rqjBJkqRud6DH\niO3cE8IAMvO7wIxmSpIkSZoeDnRG7JGIWA0MUB0j9h+A5yPi7QCZ+VBD9UmSJHWtAw1ix9T/Xr/X\n8t8DhoF3TFpFkiRJ08QBBbHMPLnpQiRJkqab/QaxiDgE+G3gwczcGBHLgAuAx4H/nJnbOlCjJElS\nVxrrYP3rgNOAlyJiEfAJ4KNUQWxVw7VJkiR1tbGC2LuBX8vM54FfB9Zm5jcy84+BtzZdnCRJUjcb\nK4i9mpm76sdLgHvH8V5JkiTtx1gH62+PiDcAvVRnTt4HEBHzAY8PkyRJOghjBbErgf8JHAb8Xmb+\nOCI+DFwNnNdwbZIk4Pzr7298G2su9ypEUgn7DWKZ+WBE/BwwNzO31IsfBxZn5nONVydJktTFxryO\nWGa+DLwcEWdSHSf2CvANwCAmSZJ0EA7ogq4RcR3wNuB/UN1j8hMRcXxmXjfK+ocCa4CjgdnANcD3\ngNuorsS/Ebg4M3dHxAXARVS3TromM9cdTEOSJElTxYGe+XgGcHJm3pSZq4CTgd/Yz/rvB36UmYuB\nXwVuBlYCK+plM4ClEXEksAxYBLwLuC4iZk+sFUmSpKnlQO81+WOqA/Z/XD+fBWzdz/pfBtbWj2dQ\nzXYtANbXy+4C3gm8CmzIzJ3AzojYBMwHvnOgDUiSJE1VY93i6FaqXYmHAE9FxJ1UoerdwN+M9r7M\nfKl+fy9VIFsB3JCZw/UqQ8A8qnA3MtDtWb5ffX1z6emZOdZqHdff31u6hFG1ubbxspf26ZY+OsHx\na2LaXt+B6pY+wF4my1gzYg/W/67fa/njY31wRPwscAfw6cz8YkT8wYiXe4EtVNci693H8v0aHNw+\n1ipFbN48VLqEferv721tbeNlL+3TqT66ZdB3/Bo/f1fax17Gv43RjBXE7snMF+uLuh6wiHg91VX4\nP5KZ36wXPxERSzLzQeB04AHgUeDaiJhDdVD/MVQH8kuSJHW9sYLYaqoD9ddT7aLc2xtHed+VQB9w\nVURcVS+7BFgVEbOAZ6juW/lqRKwCHqba/bk8M3eMswdJkqQpaawLup4REWcAp2bm9yPibOA/Ue2a\n/MR+3ncJVfDa20n7WHcAGBhX1ZIkSV1gv5eviIjLqG5nNLu+v+R/B74GvBb4g/29V5IkSfs31nXE\nfhM4KTO/B7wPuDMzVwOXUV0fTJIkSRM0VhAbzsw9p/icDNwNMOIyFJIkSZqgsQ7W3xURh1Ptivxl\nqjMhiYijqK4nJkmSpAkaa0bseuBJ4NvA6sz8YUScA3wTjxGTJEk6KGOdNbk2Iv4KeF1m/nW9+CXg\nt+rrgUmSJGmCxrzXZGa+ALww4vlfNlqRJEnSNDHWrklJkiQ1xCAmSZJUiEFMkiSpEIOYJElSIQYx\nSZKkQgxikiRJhRjEJEmSCjGISZIkFWIQkyRJKsQgJkmSVIhBTJIkqRCDmCRJUiEGMUmSpEIMYpIk\nSYUYxCRJkgoxiEmSJBViEJMkSSrEICZJklRIT5MfHhFvBT6VmUsi4peBdcBz9cufycwvRcQFwEXA\nLuCazFzXZE2SJElt0VgQi4iPAecCP6kXLQBWZuaNI9Y5ElgGHAfMAb4VEfdl5s6m6pIkSWqLJmfE\nvg+8B/h8/XwBEBGxlGpW7FJgIbChDl47I2ITMB/4ToN1SZIktUJjQSwzvxIRR49Y9CiwOjMfi4jl\nwNXAk8DWEesMAfPG+uy+vrn09MyczHInRX9/b+kSRtXm2sbLXtqnW/roBMeviWl7fQeqW/oAe5ks\njR4jtpc7MnPLnsfATcBDwMjue4Ete79xb4OD2ye/ukmwefNQ6RL2qb+/t7W1jZe9tE+n+uiWQd/x\na/z8XWkfexn/NkbTybMm74mIhfXjU4DHqGbJFkfEnIiYBxwDbOxgTZIkScV0ckbsw8BNEfEK8CJw\nYWZui4hVwMNUoXB5Zu7oYE2SJEnFNBrEMvN54IT68ePAon2sMwAMNFmHJElSG3lBV0mSpEIMYpIk\nSYUYxCRJkgoxiEmSJBViEJMkSSrEICZJklSIQUySJKkQg5gkSVIhBjFJkqRCDGKSJEmFGMQkSZIK\nMYhJkiQVYhCTJEkqxCAmSZJUiEFMkiSpEIOYJElSIQYxSZKkQgxikiRJhRjEJEmSCukpXYAkafo4\n//r7G9/Gmsvf0fg2pMnijJgkSVIhBjFJkqRCDGKSJEmFGMQkSZIKafRg/Yh4K/CpzFwSET8P3AYM\nAxuBizNzd0RcAFwE7AKuycx1TdYkSZLUFo3NiEXEx4DVwJx60UpgRWYuBmYASyPiSGAZsAh4F3Bd\nRMxuqiZJkqQ2aXLX5PeB94x4vgBYXz++CzgVWAhsyMydmbkV2ATMb7AmSZKk1mhs12RmfiUijh6x\naEZmDtePh4B5wGHA1hHr7Fm+X319c+npmTlZpU6a/v7e0iWMqs21jZe9tE+39NEJjl/Na3Mvba5t\nvOxlcnTygq67RzzuBbYA2+rHey/fr8HB7ZNb2STZvHmodAn71N/f29raxste2qdTfXTLoO/41by2\n9tItv/NgLxPZxmg6GcSeiIglmfkgcDrwAPAocG1EzAFmA8dQHcivUXTiqtTglaklSeqETgaxy4CB\niJgFPAOszcxXI2IV8DDV8WrLM3NHB2uSJEkqptEglpnPAyfUj58FTtrHOgPAQJN1SJIktZEXdJUk\nSSrEICZJklSIQUySJKkQg5gkSVIhBjFJkqRCDGKSJEmFGMQkSZIKMYhJkiQVYhCTJEkqxCAmSZJU\niEFMkiSpEIOYJElSIQYxSZKkQgxikiRJhRjEJEmSCjGISZIkFWIQkyRJKsQgJkmSVIhBTJIkqRCD\nmCRJUiEGMUmSpEIMYpIkSYUYxCRJkgoxiEmSJBViEJMkSSqkp9MbjIjHgW31078DrgVuA4aBjcDF\nmbm703VJkiR1WkeDWETMAWZk5pIRy+4EVmTmgxFxC7AUuKOTdUmSJJXQ6RmxtwBzI+LeettXAguA\n9fXrdwHvxCAmSZKmgU4Hse3ADcBq4BeogteMzByuXx8C5o31IX19c+npmdlYkRPV399buoRJ0+Ze\n2lzbeHVLL93SRyc4fjWvzb20ubbxspfJ0ekg9iywqQ5ez0bEj6hmxPboBbaM9SGDg9sbKu/gbN48\nVLqESdPWXvr7e1tb23h1Sy+d6qNbBn3Hr+a1tZdu+Z0He5nINkbT6bMmzwduBIiIfw0cBtwbEUvq\n108HHu5wTZIkSUV0ekbsT4HbIuJbVGdJng/8X2AgImYBzwBrO1yTJElSER0NYpn5MvC+fbx0Uifr\nkCRJagMv6CpJklSIQUySJKkQg5gkSVIhBjFJkqRCDGKSJEmFGMQkSZIKMYhJkiQVYhCTJEkqxCAm\nSZJUiEFMkiSpEIOYJElSIQYxSZKkQgxikiRJhRjEJEmSCjGISZIkFWIQkyRJKsQgJkmSVEhP6QI0\nfZ1//f0d2c6ay9/Rke10i058L34nklRxRkySJKkQZ8SkSeAskjT9+HuvyeCMmCRJUiEGMUmSpEIM\nYpIkSYUYxCRJkgppxcH6EXEI8GngLcBO4Lcyc1PZqiRJkprViiAGnAXMycxfiYgTgBuBpYVrkiSp\n63XTNR2n4pmsbdk1+TbgboDM/DZwXNlyJEmSmjdjeHi4dA1ExGrgK5l5V/3874E3ZuauspVJkiQ1\npy0zYtuA3hHPDzGESZKkbteWILYBeDdAfYzY02XLkSRJal5bDta/AzgtIv4KmAF8sHA9kiRJjWvF\nMWKSJEnTUVt2TUqSJE07BjFJkqRCDGKSJEmFGMQkSZIKMYjtpb7vpVosImaXruFgRcRruqSPI0rX\noH/JMazduuT33vFrEnnWJBARbwRWUt1aaRdVQH0a+GhmPluytuksIs4EbgZeAZZn5pfq5fdnZvM3\nLZtEEfGLwCeBQeALwGrgVeCSzFxXsrbxiIg37bXoc8BvAvi7Uo5jWPs4frVPW8evtlxHrLTVwBWZ\n+cieBfWFZW8FFhWrSsuBY6n+p/LliJiTmbdTXWtuqrkFuAo4GlgLvAnYAdwFTJmBDPgGsB14gep7\nCOCzwDAwpf7n0mUcw9rH8at9Wjl+GcQqc0YOYFDdfDwiStVzUCLiAWDvaeMZwHBmnligpIl6OTMH\nASJiKXB/fR/SqTiNe0hmrgfWR8TJmfmPABEx1W7ldRzVoPyZzLwvIh7IzJNLF6XuGcMcv1rJ8atB\nBrHKUxGxBrgb2Ep138t3A39dtKqJuxwYAM6m2k0xVT0fESuBqzJzKCLeA9wDHF64ronI+ub2F2bm\neQARcTnwYtGqxikz/zEizgFuiIjjS9ejf9ZNY5jjV/s4fjXIIFb5beAs4G3AYVQ3IV9HdeulKScz\nH4mIzwPzM3NK9lA7H3g/9V+QmfmDiDgZuKJoVRNzAXBmZu4esez/AKsK1TNhmbkLuDQizsMTftqi\na8Ywx69WcvxqkAfrS5IkFdKKNChJkjQdGcQkSZIKMYhJkiQVYhCTJEkq5P8BjG6axKSlfFcAAAAA\nSUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c40527a5c0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmIAAAE4CAYAAAD4q5/qAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c404b2e0b8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmIAAAE4CAYAAAD4q5/qAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c4048bea90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmIAAAE4CAYAAAD4q5/qAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nJBViEJMkSSrEICZJklSIQUySJKmQ/wZXRAp4AmmOUwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c40738b6a0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmIAAAE+CAYAAAAu8nz3AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c404b0f630>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmIAAAE4CAYAAAD4q5/qAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n4q3wezeohAcAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c404fcfac8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmIAAAE4CAYAAAD4q5/qAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nUiEGMUmSpEL+P4fACCpPF7BMAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c404d89d68>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "## Visualize barplot group by survived\n",
    "\n",
    "print('dead\\n', train_set.loc[train_set['Survived'] == 0, 'Sex_Numeric'].value_counts(sort = False))\n",
    "print('survived\\n', train_set.loc[train_set['Survived'] == 1, 'Sex_Numeric'].value_counts(sort = False))\n",
    "\n",
    "for col in set(list(train_set)) - set(['Fare']):\n",
    "    \n",
    "    fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(10,5), sharex = True, sharey = True)\n",
    "    train_set.loc[train_set['Survived'] == 0, col].value_counts(sort = False).plot(kind = 'bar', ax = ax1)\n",
    "    ax1.set_title('Dead')\n",
    "    train_set.loc[train_set['Survived'] == 1, col].value_counts(sort = False).plot(kind = 'bar', ax = ax2)\n",
    "    ax2.set_title('Survived')\n",
    "    ax1.set_ylabel(col)\n",
    "    \n",
    "print(train_set.loc[train_set['Survived'] == 0, 'Survived'].value_counts())\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 222,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0    549\n",
      "Name: Survived, dtype: int64\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAl8AAAE4CAYAAABlppZiAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c407671b38>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAl8AAAE4CAYAAABlppZiAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nSSro/wPwAGo70N+LnAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c406b4ae10>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAl8AAAE4CAYAAABlppZiAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c404daab70>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAl8AAAE4CAYAAABlppZiAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nSZIkFWT4kiRJKsjwJUmSVND/Ailn1zHYt0VnAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c407645f28>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAl8AAAE4CAYAAABlppZiAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n8CVJklSQ4UuSJKmg/w9RpUWmsGIRwwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c406b48c88>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAl8AAAE+CAYAAACz/3V/AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nkhpk+JIkSWrQfwPMfCYDrWoVcAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c40549cc88>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAl8AAAE4CAYAAABlppZiAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n+JIkSaqR4UuSJKlGhi9JkqQa/X/kkTofoi3pHQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c406dce7b8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAl8AAAE4CAYAAABlppZiAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c4001fc940>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "## Visualize barplot group by survived and divide by number of survival \n",
    "# to observe each category in normalize way\n",
    "\n",
    "num_survive = len(train_set.loc[train_set['Survived'] == 1, 'Survived'])\n",
    "num_notsurvive = len(train_set.loc[train_set['Survived'] == 0, 'Survived'])\n",
    "for col in set(list(train_set)) - set(['Fare']):\n",
    "    \n",
    "    fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(10,5), sharex = True, sharey = True)\n",
    "    (train_set.loc[train_set['Survived'] == 0, col].value_counts(sort = False)/num_notsurvive).plot(kind = 'bar', ax = ax1)\n",
    "    ax1.set_title('Dead')\n",
    "    (train_set.loc[train_set['Survived'] == 1, col].value_counts(sort = False)/num_survive).plot(kind = 'bar', ax = ax2)\n",
    "    ax2.set_title('Survived')\n",
    "    ax1.set_ylabel(col + ' Ratio')\n",
    "    \n",
    "print(train_set.loc[train_set['Survived'] == 0, 'Survived'].value_counts())\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Transform Categorical data to indicator data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 223,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "     Survived      Fare  Pclass_1  Pclass_2  Pclass_3  Name_Numeric_0  \\\n",
      "0           0    7.2500         0         0         1               1   \n",
      "1           1   71.2833         1         0         0               0   \n",
      "2           1    7.9250         0         0         1               0   \n",
      "3           1   53.1000         1         0         0               0   \n",
      "4           0    8.0500         0         0         1               1   \n",
      "5           0    8.4583         0         0         1               1   \n",
      "6           0   51.8625         1         0         0               1   \n",
      "7           0   21.0750         0         0         1               0   \n",
      "8           1   11.1333         0         0         1               0   \n",
      "9           1   30.0708         0         1         0               0   \n",
      "10          1   16.7000         0         0         1               0   \n",
      "11          1   26.5500         1         0         0               0   \n",
      "12          0    8.0500         0         0         1               1   \n",
      "13          0   31.2750         0         0         1               1   \n",
      "14          0    7.8542         0         0         1               0   \n",
      "15          1   16.0000         0         1         0               0   \n",
      "16          0   29.1250         0         0         1               0   \n",
      "17          1   13.0000         0         1         0               1   \n",
      "18          0   18.0000         0         0         1               0   \n",
      "19          1    7.2250         0         0         1               0   \n",
      "20          0   26.0000         0         1         0               1   \n",
      "21          1   13.0000         0         1         0               1   \n",
      "22          1    8.0292         0         0         1               0   \n",
      "23          1   35.5000         1         0         0               1   \n",
      "24          0   21.0750         0         0         1               0   \n",
      "25          1   31.3875         0         0         1               0   \n",
      "26          0    7.2250         0         0         1               1   \n",
      "27          0  263.0000         1         0         0               1   \n",
      "28          1    7.8792         0         0         1               0   \n",
      "29          0    7.8958         0         0         1               1   \n",
      "..        ...       ...       ...       ...       ...             ...   \n",
      "861         0   11.5000         0         1         0               1   \n",
      "862         1   25.9292         1         0         0               0   \n",
      "863         0   69.5500         0         0         1               0   \n",
      "864         0   13.0000         0         1         0               1   \n",
      "865         1   13.0000         0         1         0               0   \n",
      "866         1   13.8583         0         1         0               0   \n",
      "867         0   50.4958         1         0         0               1   \n",
      "868         0    9.5000         0         0         1               1   \n",
      "869         1   11.1333         0         0         1               0   \n",
      "870         0    7.8958         0         0         1               1   \n",
      "871         1   52.5542         1         0         0               0   \n",
      "872         0    5.0000         1         0         0               1   \n",
      "873         0    9.0000         0         0         1               1   \n",
      "874         1   24.0000         0         1         0               0   \n",
      "875         1    7.2250         0         0         1               0   \n",
      "876         0    9.8458         0         0         1               1   \n",
      "877         0    7.8958         0         0         1               1   \n",
      "878         0    7.8958         0         0         1               1   \n",
      "879         1   83.1583         1         0         0               0   \n",
      "880         1   26.0000         0         1         0               0   \n",
      "881         0    7.8958         0         0         1               1   \n",
      "882         0   10.5167         0         0         1               0   \n",
      "883         0   10.5000         0         1         0               1   \n",
      "884         0    7.0500         0         0         1               1   \n",
      "885         0   29.1250         0         0         1               0   \n",
      "886         0   13.0000         0         1         0               0   \n",
      "887         1   30.0000         1         0         0               0   \n",
      "888         0   23.4500         0         0         1               0   \n",
      "889         1   30.0000         1         0         0               1   \n",
      "890         0    7.7500         0         0         1               1   \n",
      "\n",
      "     Name_Numeric_1  Name_Numeric_2  Name_Numeric_3  Name_Numeric_4   ...     \\\n",
      "0                 0               0               0               0   ...      \n",
      "1                 1               0               0               0   ...      \n",
      "2                 0               1               0               0   ...      \n",
      "3                 1               0               0               0   ...      \n",
      "4                 0               0               0               0   ...      \n",
      "5                 0               0               0               0   ...      \n",
      "6                 0               0               0               0   ...      \n",
      "7                 0               0               1               0   ...      \n",
      "8                 1               0               0               0   ...      \n",
      "9                 1               0               0               0   ...      \n",
      "10                0               1               0               0   ...      \n",
      "11                0               1               0               0   ...      \n",
      "12                0               0               0               0   ...      \n",
      "13                0               0               0               0   ...      \n",
      "14                0               1               0               0   ...      \n",
      "15                1               0               0               0   ...      \n",
      "16                0               0               1               0   ...      \n",
      "17                0               0               0               0   ...      \n",
      "18                1               0               0               0   ...      \n",
      "19                1               0               0               0   ...      \n",
      "20                0               0               0               0   ...      \n",
      "21                0               0               0               0   ...      \n",
      "22                0               1               0               0   ...      \n",
      "23                0               0               0               0   ...      \n",
      "24                0               1               0               0   ...      \n",
      "25                1               0               0               0   ...      \n",
      "26                0               0               0               0   ...      \n",
      "27                0               0               0               0   ...      \n",
      "28                0               1               0               0   ...      \n",
      "29                0               0               0               0   ...      \n",
      "..              ...             ...             ...             ...   ...      \n",
      "861               0               0               0               0   ...      \n",
      "862               1               0               0               0   ...      \n",
      "863               0               1               0               0   ...      \n",
      "864               0               0               0               0   ...      \n",
      "865               1               0               0               0   ...      \n",
      "866               0               1               0               0   ...      \n",
      "867               0               0               0               0   ...      \n",
      "868               0               0               0               0   ...      \n",
      "869               0               0               1               0   ...      \n",
      "870               0               0               0               0   ...      \n",
      "871               1               0               0               0   ...      \n",
      "872               0               0               0               0   ...      \n",
      "873               0               0               0               0   ...      \n",
      "874               1               0               0               0   ...      \n",
      "875               0               1               0               0   ...      \n",
      "876               0               0               0               0   ...      \n",
      "877               0               0               0               0   ...      \n",
      "878               0               0               0               0   ...      \n",
      "879               1               0               0               0   ...      \n",
      "880               1               0               0               0   ...      \n",
      "881               0               0               0               0   ...      \n",
      "882               0               1               0               0   ...      \n",
      "883               0               0               0               0   ...      \n",
      "884               0               0               0               0   ...      \n",
      "885               1               0               0               0   ...      \n",
      "886               0               0               0               0   ...      \n",
      "887               0               1               0               0   ...      \n",
      "888               0               1               0               0   ...      \n",
      "889               0               0               0               0   ...      \n",
      "890               0               0               0               0   ...      \n",
      "\n",
      "     Cabin_Numeric_0  Cabin_Numeric_1  Cabin_Numeric_2  Cabin_Numeric_3  \\\n",
      "0                  1                0                0                0   \n",
      "1                  0                1                0                0   \n",
      "2                  1                0                0                0   \n",
      "3                  0                1                0                0   \n",
      "4                  1                0                0                0   \n",
      "5                  1                0                0                0   \n",
      "6                  0                0                1                0   \n",
      "7                  1                0                0                0   \n",
      "8                  1                0                0                0   \n",
      "9                  1                0                0                0   \n",
      "10                 0                0                0                1   \n",
      "11                 0                1                0                0   \n",
      "12                 1                0                0                0   \n",
      "13                 1                0                0                0   \n",
      "14                 1                0                0                0   \n",
      "15                 1                0                0                0   \n",
      "16                 1                0                0                0   \n",
      "17                 1                0                0                0   \n",
      "18                 1                0                0                0   \n",
      "19                 1                0                0                0   \n",
      "20                 1                0                0                0   \n",
      "21                 0                0                0                0   \n",
      "22                 1                0                0                0   \n",
      "23                 0                0                0                0   \n",
      "24                 1                0                0                0   \n",
      "25                 1                0                0                0   \n",
      "26                 1                0                0                0   \n",
      "27                 0                1                0                0   \n",
      "28                 1                0                0                0   \n",
      "29                 1                0                0                0   \n",
      "..               ...              ...              ...              ...   \n",
      "861                1                0                0                0   \n",
      "862                0                0                0                0   \n",
      "863                1                0                0                0   \n",
      "864                1                0                0                0   \n",
      "865                1                0                0                0   \n",
      "866                1                0                0                0   \n",
      "867                0                0                0                0   \n",
      "868                1                0                0                0   \n",
      "869                1                0                0                0   \n",
      "870                1                0                0                0   \n",
      "871                0                0                0                0   \n",
      "872                0                0                0                0   \n",
      "873                1                0                0                0   \n",
      "874                1                0                0                0   \n",
      "875                1                0                0                0   \n",
      "876                1                0                0                0   \n",
      "877                1                0                0                0   \n",
      "878                1                0                0                0   \n",
      "879                0                1                0                0   \n",
      "880                1                0                0                0   \n",
      "881                1                0                0                0   \n",
      "882                1                0                0                0   \n",
      "883                1                0                0                0   \n",
      "884                1                0                0                0   \n",
      "885                1                0                0                0   \n",
      "886                1                0                0                0   \n",
      "887                0                0                0                0   \n",
      "888                1                0                0                0   \n",
      "889                0                1                0                0   \n",
      "890                1                0                0                0   \n",
      "\n",
      "     Cabin_Numeric_4  Cabin_Numeric_5  Cabin_Numeric_6  Cabin_Numeric_7  \\\n",
      "0                  0                0                0                0   \n",
      "1                  0                0                0                0   \n",
      "2                  0                0                0                0   \n",
      "3                  0                0                0                0   \n",
      "4                  0                0                0                0   \n",
      "5                  0                0                0                0   \n",
      "6                  0                0                0                0   \n",
      "7                  0                0                0                0   \n",
      "8                  0                0                0                0   \n",
      "9                  0                0                0                0   \n",
      "10                 0                0                0                0   \n",
      "11                 0                0                0                0   \n",
      "12                 0                0                0                0   \n",
      "13                 0                0                0                0   \n",
      "14                 0                0                0                0   \n",
      "15                 0                0                0                0   \n",
      "16                 0                0                0                0   \n",
      "17                 0                0                0                0   \n",
      "18                 0                0                0                0   \n",
      "19                 0                0                0                0   \n",
      "20                 0                0                0                0   \n",
      "21                 1                0                0                0   \n",
      "22                 0                0                0                0   \n",
      "23                 0                1                0                0   \n",
      "24                 0                0                0                0   \n",
      "25                 0                0                0                0   \n",
      "26                 0                0                0                0   \n",
      "27                 0                0                0                0   \n",
      "28                 0                0                0                0   \n",
      "29                 0                0                0                0   \n",
      "..               ...              ...              ...              ...   \n",
      "861                0                0                0                0   \n",
      "862                1                0                0                0   \n",
      "863                0                0                0                0   \n",
      "864                0                0                0                0   \n",
      "865                0                0                0                0   \n",
      "866                0                0                0                0   \n",
      "867                0                1                0                0   \n",
      "868                0                0                0                0   \n",
      "869                0                0                0                0   \n",
      "870                0                0                0                0   \n",
      "871                1                0                0                0   \n",
      "872                0                0                1                0   \n",
      "873                0                0                0                0   \n",
      "874                0                0                0                0   \n",
      "875                0                0                0                0   \n",
      "876                0                0                0                0   \n",
      "877                0                0                0                0   \n",
      "878                0                0                0                0   \n",
      "879                0                0                0                0   \n",
      "880                0                0                0                0   \n",
      "881                0                0                0                0   \n",
      "882                0                0                0                0   \n",
      "883                0                0                0                0   \n",
      "884                0                0                0                0   \n",
      "885                0                0                0                0   \n",
      "886                0                0                0                0   \n",
      "887                0                0                1                0   \n",
      "888                0                0                0                0   \n",
      "889                0                0                0                0   \n",
      "890                0                0                0                0   \n",
      "\n",
      "     Cabin_Numeric_8  Parch_9  \n",
      "0                  0      0.0  \n",
      "1                  0      0.0  \n",
      "2                  0      0.0  \n",
      "3                  0      0.0  \n",
      "4                  0      0.0  \n",
      "5                  0      0.0  \n",
      "6                  0      0.0  \n",
      "7                  0      0.0  \n",
      "8                  0      0.0  \n",
      "9                  0      0.0  \n",
      "10                 0      0.0  \n",
      "11                 0      0.0  \n",
      "12                 0      0.0  \n",
      "13                 0      0.0  \n",
      "14                 0      0.0  \n",
      "15                 0      0.0  \n",
      "16                 0      0.0  \n",
      "17                 0      0.0  \n",
      "18                 0      0.0  \n",
      "19                 0      0.0  \n",
      "20                 0      0.0  \n",
      "21                 0      0.0  \n",
      "22                 0      0.0  \n",
      "23                 0      0.0  \n",
      "24                 0      0.0  \n",
      "25                 0      0.0  \n",
      "26                 0      0.0  \n",
      "27                 0      0.0  \n",
      "28                 0      0.0  \n",
      "29                 0      0.0  \n",
      "..               ...      ...  \n",
      "861                0      0.0  \n",
      "862                0      0.0  \n",
      "863                0      0.0  \n",
      "864                0      0.0  \n",
      "865                0      0.0  \n",
      "866                0      0.0  \n",
      "867                0      0.0  \n",
      "868                0      0.0  \n",
      "869                0      0.0  \n",
      "870                0      0.0  \n",
      "871                0      0.0  \n",
      "872                0      0.0  \n",
      "873                0      0.0  \n",
      "874                0      0.0  \n",
      "875                0      0.0  \n",
      "876                0      0.0  \n",
      "877                0      0.0  \n",
      "878                0      0.0  \n",
      "879                0      0.0  \n",
      "880                0      0.0  \n",
      "881                0      0.0  \n",
      "882                0      0.0  \n",
      "883                0      0.0  \n",
      "884                0      0.0  \n",
      "885                0      0.0  \n",
      "886                0      0.0  \n",
      "887                0      0.0  \n",
      "888                0      0.0  \n",
      "889                0      0.0  \n",
      "890                0      0.0  \n",
      "\n",
      "[891 rows x 52 columns]\n",
      "         Fare  Pclass_1  Pclass_2  Pclass_3  Name_Numeric_0  Name_Numeric_1  \\\n",
      "0      7.8292         0         0         1               1               0   \n",
      "1      7.0000         0         0         1               0               1   \n",
      "2      9.6875         0         1         0               1               0   \n",
      "3      8.6625         0         0         1               1               0   \n",
      "4     12.2875         0         0         1               0               1   \n",
      "5      9.2250         0         0         1               1               0   \n",
      "6      7.6292         0         0         1               0               0   \n",
      "7     29.0000         0         1         0               1               0   \n",
      "8      7.2292         0         0         1               0               1   \n",
      "9     24.1500         0         0         1               1               0   \n",
      "10     7.8958         0         0         1               1               0   \n",
      "11    26.0000         1         0         0               1               0   \n",
      "12    82.2667         1         0         0               0               1   \n",
      "13    26.0000         0         1         0               1               0   \n",
      "14    61.1750         1         0         0               0               1   \n",
      "15    27.7208         0         1         0               0               1   \n",
      "16    12.3500         0         1         0               1               0   \n",
      "17     7.2250         0         0         1               1               0   \n",
      "18     7.9250         0         0         1               0               0   \n",
      "19     7.2250         0         0         1               0               1   \n",
      "20    59.4000         1         0         0               1               0   \n",
      "21     3.1708         0         0         1               0               0   \n",
      "22    31.6833         1         0         0               0               1   \n",
      "23    61.3792         1         0         0               1               0   \n",
      "24   262.3750         1         0         0               0               1   \n",
      "25    14.5000         0         0         1               1               0   \n",
      "26    61.9792         1         0         0               0               0   \n",
      "27     7.2250         0         0         1               1               0   \n",
      "28    30.5000         1         0         0               1               0   \n",
      "29    21.6792         0         0         1               1               0   \n",
      "..        ...       ...       ...       ...             ...             ...   \n",
      "388    7.7500         0         0         1               1               0   \n",
      "389   21.0750         0         0         1               0               0   \n",
      "390   93.5000         1         0         0               1               0   \n",
      "391   39.4000         1         0         0               0               1   \n",
      "392   20.2500         0         0         1               0               0   \n",
      "393   10.5000         0         1         0               1               0   \n",
      "394   22.0250         0         0         1               1               0   \n",
      "395   60.0000         1         0         0               0               1   \n",
      "396    7.2500         0         0         1               1               0   \n",
      "397   79.2000         1         0         0               0               1   \n",
      "398    7.7750         0         0         1               1               0   \n",
      "399    7.7333         0         0         1               1               0   \n",
      "400  164.8667         1         0         0               0               0   \n",
      "401   21.0000         0         1         0               1               0   \n",
      "402   59.4000         1         0         0               0               0   \n",
      "403   47.1000         1         0         0               1               0   \n",
      "404   27.7208         1         0         0               1               0   \n",
      "405   13.8625         0         1         0               1               0   \n",
      "406   10.5000         0         1         0               1               0   \n",
      "407  211.5000         1         0         0               1               0   \n",
      "408    7.7208         0         0         1               0               0   \n",
      "409   13.7750         0         0         1               0               0   \n",
      "410    7.7500         0         0         1               0               0   \n",
      "411   90.0000         1         0         0               0               1   \n",
      "412    7.7750         0         0         1               0               0   \n",
      "413    8.0500         0         0         1               1               0   \n",
      "414  108.9000         1         0         0               1               0   \n",
      "415    7.2500         0         0         1               1               0   \n",
      "416    8.0500         0         0         1               1               0   \n",
      "417   22.3583         0         0         1               0               0   \n",
      "\n",
      "     Name_Numeric_2  Name_Numeric_3  Name_Numeric_5  Name_Numeric_6  \\\n",
      "0                 0               0               0               0   \n",
      "1                 0               0               0               0   \n",
      "2                 0               0               0               0   \n",
      "3                 0               0               0               0   \n",
      "4                 0               0               0               0   \n",
      "5                 0               0               0               0   \n",
      "6                 1               0               0               0   \n",
      "7                 0               0               0               0   \n",
      "8                 0               0               0               0   \n",
      "9                 0               0               0               0   \n",
      "10                0               0               0               0   \n",
      "11                0               0               0               0   \n",
      "12                0               0               0               0   \n",
      "13                0               0               0               0   \n",
      "14                0               0               0               0   \n",
      "15                0               0               0               0   \n",
      "16                0               0               0               0   \n",
      "17                0               0               0               0   \n",
      "18                1               0               0               0   \n",
      "19                0               0               0               0   \n",
      "20                0               0               0               0   \n",
      "21                0               1               0               0   \n",
      "22                0               0               0               0   \n",
      "23                0               0               0               0   \n",
      "24                0               0               0               0   \n",
      "25                0               0               0               0   \n",
      "26                1               0               0               0   \n",
      "27                0               0               0               0   \n",
      "28                0               0               0               0   \n",
      "29                0               0               0               0   \n",
      "..              ...             ...             ...             ...   \n",
      "388               0               0               0               0   \n",
      "389               0               1               0               0   \n",
      "390               0               0               0               0   \n",
      "391               0               0               0               0   \n",
      "392               0               1               0               0   \n",
      "393               0               0               0               0   \n",
      "394               0               0               0               0   \n",
      "395               0               0               0               0   \n",
      "396               0               0               0               0   \n",
      "397               0               0               0               0   \n",
      "398               0               0               0               0   \n",
      "399               0               0               0               0   \n",
      "400               1               0               0               0   \n",
      "401               0               0               0               0   \n",
      "402               1               0               0               0   \n",
      "403               0               0               0               0   \n",
      "404               0               0               0               0   \n",
      "405               0               0               0               0   \n",
      "406               0               0               0               0   \n",
      "407               0               0               0               0   \n",
      "408               1               0               0               0   \n",
      "409               1               0               0               0   \n",
      "410               1               0               0               0   \n",
      "411               0               0               0               0   \n",
      "412               1               0               0               0   \n",
      "413               0               0               0               0   \n",
      "414               0               0               0               0   \n",
      "415               0               0               0               0   \n",
      "416               0               0               0               0   \n",
      "417               0               1               0               0   \n",
      "\n",
      "          ...         Embarked_Numeric_3  Cabin_Numeric_8  Name_Numeric_12  \\\n",
      "0         ...                        0.0              0.0              0.0   \n",
      "1         ...                        0.0              0.0              0.0   \n",
      "2         ...                        0.0              0.0              0.0   \n",
      "3         ...                        0.0              0.0              0.0   \n",
      "4         ...                        0.0              0.0              0.0   \n",
      "5         ...                        0.0              0.0              0.0   \n",
      "6         ...                        0.0              0.0              0.0   \n",
      "7         ...                        0.0              0.0              0.0   \n",
      "8         ...                        0.0              0.0              0.0   \n",
      "9         ...                        0.0              0.0              0.0   \n",
      "10        ...                        0.0              0.0              0.0   \n",
      "11        ...                        0.0              0.0              0.0   \n",
      "12        ...                        0.0              0.0              0.0   \n",
      "13        ...                        0.0              0.0              0.0   \n",
      "14        ...                        0.0              0.0              0.0   \n",
      "15        ...                        0.0              0.0              0.0   \n",
      "16        ...                        0.0              0.0              0.0   \n",
      "17        ...                        0.0              0.0              0.0   \n",
      "18        ...                        0.0              0.0              0.0   \n",
      "19        ...                        0.0              0.0              0.0   \n",
      "20        ...                        0.0              0.0              0.0   \n",
      "21        ...                        0.0              0.0              0.0   \n",
      "22        ...                        0.0              0.0              0.0   \n",
      "23        ...                        0.0              0.0              0.0   \n",
      "24        ...                        0.0              0.0              0.0   \n",
      "25        ...                        0.0              0.0              0.0   \n",
      "26        ...                        0.0              0.0              0.0   \n",
      "27        ...                        0.0              0.0              0.0   \n",
      "28        ...                        0.0              0.0              0.0   \n",
      "29        ...                        0.0              0.0              0.0   \n",
      "..        ...                        ...              ...              ...   \n",
      "388       ...                        0.0              0.0              0.0   \n",
      "389       ...                        0.0              0.0              0.0   \n",
      "390       ...                        0.0              0.0              0.0   \n",
      "391       ...                        0.0              0.0              0.0   \n",
      "392       ...                        0.0              0.0              0.0   \n",
      "393       ...                        0.0              0.0              0.0   \n",
      "394       ...                        0.0              0.0              0.0   \n",
      "395       ...                        0.0              0.0              0.0   \n",
      "396       ...                        0.0              0.0              0.0   \n",
      "397       ...                        0.0              0.0              0.0   \n",
      "398       ...                        0.0              0.0              0.0   \n",
      "399       ...                        0.0              0.0              0.0   \n",
      "400       ...                        0.0              0.0              0.0   \n",
      "401       ...                        0.0              0.0              0.0   \n",
      "402       ...                        0.0              0.0              0.0   \n",
      "403       ...                        0.0              0.0              0.0   \n",
      "404       ...                        0.0              0.0              0.0   \n",
      "405       ...                        0.0              0.0              0.0   \n",
      "406       ...                        0.0              0.0              0.0   \n",
      "407       ...                        0.0              0.0              0.0   \n",
      "408       ...                        0.0              0.0              0.0   \n",
      "409       ...                        0.0              0.0              0.0   \n",
      "410       ...                        0.0              0.0              0.0   \n",
      "411       ...                        0.0              0.0              0.0   \n",
      "412       ...                        0.0              0.0              0.0   \n",
      "413       ...                        0.0              0.0              0.0   \n",
      "414       ...                        0.0              0.0              0.0   \n",
      "415       ...                        0.0              0.0              0.0   \n",
      "416       ...                        0.0              0.0              0.0   \n",
      "417       ...                        0.0              0.0              0.0   \n",
      "\n",
      "     Name_Numeric_14  Name_Numeric_9  Name_Numeric_10  Name_Numeric_7  \\\n",
      "0                0.0             0.0              0.0             0.0   \n",
      "1                0.0             0.0              0.0             0.0   \n",
      "2                0.0             0.0              0.0             0.0   \n",
      "3                0.0             0.0              0.0             0.0   \n",
      "4                0.0             0.0              0.0             0.0   \n",
      "5                0.0             0.0              0.0             0.0   \n",
      "6                0.0             0.0              0.0             0.0   \n",
      "7                0.0             0.0              0.0             0.0   \n",
      "8                0.0             0.0              0.0             0.0   \n",
      "9                0.0             0.0              0.0             0.0   \n",
      "10               0.0             0.0              0.0             0.0   \n",
      "11               0.0             0.0              0.0             0.0   \n",
      "12               0.0             0.0              0.0             0.0   \n",
      "13               0.0             0.0              0.0             0.0   \n",
      "14               0.0             0.0              0.0             0.0   \n",
      "15               0.0             0.0              0.0             0.0   \n",
      "16               0.0             0.0              0.0             0.0   \n",
      "17               0.0             0.0              0.0             0.0   \n",
      "18               0.0             0.0              0.0             0.0   \n",
      "19               0.0             0.0              0.0             0.0   \n",
      "20               0.0             0.0              0.0             0.0   \n",
      "21               0.0             0.0              0.0             0.0   \n",
      "22               0.0             0.0              0.0             0.0   \n",
      "23               0.0             0.0              0.0             0.0   \n",
      "24               0.0             0.0              0.0             0.0   \n",
      "25               0.0             0.0              0.0             0.0   \n",
      "26               0.0             0.0              0.0             0.0   \n",
      "27               0.0             0.0              0.0             0.0   \n",
      "28               0.0             0.0              0.0             0.0   \n",
      "29               0.0             0.0              0.0             0.0   \n",
      "..               ...             ...              ...             ...   \n",
      "388              0.0             0.0              0.0             0.0   \n",
      "389              0.0             0.0              0.0             0.0   \n",
      "390              0.0             0.0              0.0             0.0   \n",
      "391              0.0             0.0              0.0             0.0   \n",
      "392              0.0             0.0              0.0             0.0   \n",
      "393              0.0             0.0              0.0             0.0   \n",
      "394              0.0             0.0              0.0             0.0   \n",
      "395              0.0             0.0              0.0             0.0   \n",
      "396              0.0             0.0              0.0             0.0   \n",
      "397              0.0             0.0              0.0             0.0   \n",
      "398              0.0             0.0              0.0             0.0   \n",
      "399              0.0             0.0              0.0             0.0   \n",
      "400              0.0             0.0              0.0             0.0   \n",
      "401              0.0             0.0              0.0             0.0   \n",
      "402              0.0             0.0              0.0             0.0   \n",
      "403              0.0             0.0              0.0             0.0   \n",
      "404              0.0             0.0              0.0             0.0   \n",
      "405              0.0             0.0              0.0             0.0   \n",
      "406              0.0             0.0              0.0             0.0   \n",
      "407              0.0             0.0              0.0             0.0   \n",
      "408              0.0             0.0              0.0             0.0   \n",
      "409              0.0             0.0              0.0             0.0   \n",
      "410              0.0             0.0              0.0             0.0   \n",
      "411              0.0             0.0              0.0             0.0   \n",
      "412              0.0             0.0              0.0             0.0   \n",
      "413              0.0             0.0              0.0             0.0   \n",
      "414              0.0             0.0              0.0             0.0   \n",
      "415              0.0             0.0              0.0             0.0   \n",
      "416              0.0             0.0              0.0             0.0   \n",
      "417              0.0             0.0              0.0             0.0   \n",
      "\n",
      "     Name_Numeric_15  Name_Numeric_4  Name_Numeric_16  \n",
      "0                0.0             0.0              0.0  \n",
      "1                0.0             0.0              0.0  \n",
      "2                0.0             0.0              0.0  \n",
      "3                0.0             0.0              0.0  \n",
      "4                0.0             0.0              0.0  \n",
      "5                0.0             0.0              0.0  \n",
      "6                0.0             0.0              0.0  \n",
      "7                0.0             0.0              0.0  \n",
      "8                0.0             0.0              0.0  \n",
      "9                0.0             0.0              0.0  \n",
      "10               0.0             0.0              0.0  \n",
      "11               0.0             0.0              0.0  \n",
      "12               0.0             0.0              0.0  \n",
      "13               0.0             0.0              0.0  \n",
      "14               0.0             0.0              0.0  \n",
      "15               0.0             0.0              0.0  \n",
      "16               0.0             0.0              0.0  \n",
      "17               0.0             0.0              0.0  \n",
      "18               0.0             0.0              0.0  \n",
      "19               0.0             0.0              0.0  \n",
      "20               0.0             0.0              0.0  \n",
      "21               0.0             0.0              0.0  \n",
      "22               0.0             0.0              0.0  \n",
      "23               0.0             0.0              0.0  \n",
      "24               0.0             0.0              0.0  \n",
      "25               0.0             0.0              0.0  \n",
      "26               0.0             0.0              0.0  \n",
      "27               0.0             0.0              0.0  \n",
      "28               0.0             0.0              0.0  \n",
      "29               0.0             0.0              0.0  \n",
      "..               ...             ...              ...  \n",
      "388              0.0             0.0              0.0  \n",
      "389              0.0             0.0              0.0  \n",
      "390              0.0             0.0              0.0  \n",
      "391              0.0             0.0              0.0  \n",
      "392              0.0             0.0              0.0  \n",
      "393              0.0             0.0              0.0  \n",
      "394              0.0             0.0              0.0  \n",
      "395              0.0             0.0              0.0  \n",
      "396              0.0             0.0              0.0  \n",
      "397              0.0             0.0              0.0  \n",
      "398              0.0             0.0              0.0  \n",
      "399              0.0             0.0              0.0  \n",
      "400              0.0             0.0              0.0  \n",
      "401              0.0             0.0              0.0  \n",
      "402              0.0             0.0              0.0  \n",
      "403              0.0             0.0              0.0  \n",
      "404              0.0             0.0              0.0  \n",
      "405              0.0             0.0              0.0  \n",
      "406              0.0             0.0              0.0  \n",
      "407              0.0             0.0              0.0  \n",
      "408              0.0             0.0              0.0  \n",
      "409              0.0             0.0              0.0  \n",
      "410              0.0             0.0              0.0  \n",
      "411              0.0             0.0              0.0  \n",
      "412              0.0             0.0              0.0  \n",
      "413              0.0             0.0              0.0  \n",
      "414              0.0             0.0              0.0  \n",
      "415              0.0             0.0              0.0  \n",
      "416              0.0             0.0              0.0  \n",
      "417              0.0             0.0              0.0  \n",
      "\n",
      "[418 rows x 51 columns]\n"
     ]
    }
   ],
   "source": [
    "## Transform All categorical data to indicator data\n",
    "\n",
    "train_data_copy = train_set.copy()\n",
    "test_data_copy = test_set.copy()\n",
    "for col in set(list(train_set)) - set(['Fare', 'Survived']):\n",
    "    train_data_copy = pd.get_dummies(train_data_copy, columns = [col])\n",
    "    test_data_copy = pd.get_dummies(test_data_copy, columns = [col])\n",
    "\n",
    "col_train = list(train_data_copy)\n",
    "col_test = list(test_data_copy)\n",
    "    \n",
    "for col in set(col_train) - set(['Survived']):\n",
    "    if col not in col_test:\n",
    "        test_data_copy[col] = np.zeros(test_data_copy[col_test[0]].shape[0])\n",
    "\n",
    "for col in list(test_data_copy):\n",
    "    if col not in list(train_data_copy):\n",
    "        train_data_copy[col] = np.zeros(train_data_copy[col_train[0]].shape[0])\n",
    "print(train_data_copy)\n",
    "print(test_data_copy)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Log transformation "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 224,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fare                  0\n",
      "Pclass_1              0\n",
      "Pclass_2              0\n",
      "Pclass_3              0\n",
      "Name_Numeric_0        0\n",
      "Name_Numeric_1        0\n",
      "Name_Numeric_2        0\n",
      "Name_Numeric_3        0\n",
      "Name_Numeric_5        0\n",
      "Name_Numeric_6        0\n",
      "Name_Numeric_8        0\n",
      "Name_Numeric_13       0\n",
      "Sex_Numeric_0         0\n",
      "Sex_Numeric_1         0\n",
      "SibSp_0               0\n",
      "SibSp_1               0\n",
      "SibSp_2               0\n",
      "SibSp_3               0\n",
      "SibSp_4               0\n",
      "SibSp_5               0\n",
      "SibSp_8               0\n",
      "Parch_0               0\n",
      "Parch_1               0\n",
      "Parch_2               0\n",
      "Parch_3               0\n",
      "Parch_4               0\n",
      "Parch_5               0\n",
      "Parch_6               0\n",
      "Parch_9               0\n",
      "Embarked_Numeric_0    0\n",
      "Embarked_Numeric_1    0\n",
      "Embarked_Numeric_2    0\n",
      "Cabin_Numeric_0       0\n",
      "Cabin_Numeric_1       0\n",
      "Cabin_Numeric_2       0\n",
      "Cabin_Numeric_3       0\n",
      "Cabin_Numeric_4       0\n",
      "Cabin_Numeric_5       0\n",
      "Cabin_Numeric_6       0\n",
      "Cabin_Numeric_7       0\n",
      "Name_Numeric_11       0\n",
      "Embarked_Numeric_3    0\n",
      "Cabin_Numeric_8       0\n",
      "Name_Numeric_12       0\n",
      "Name_Numeric_14       0\n",
      "Name_Numeric_9        0\n",
      "Name_Numeric_10       0\n",
      "Name_Numeric_7        0\n",
      "Name_Numeric_15       0\n",
      "Name_Numeric_4        0\n",
      "Name_Numeric_16       0\n",
      "dtype: int64\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "'\\nplot_train = train_data_copy.copy()\\nplot_test = test_data_copy.copy()\\n'"
      ]
     },
     "execution_count": 224,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYMAAAD4CAYAAAAO9oqkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nhoEkCcNAkgT8HyLHdjFpzn1aAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2c4053e4d68>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "## Transform exponential like distribution of Fare to Normal distribution\n",
    "## using box-cox transformation\n",
    "\n",
    "#train_data_copy.loc[np.logical_not(train_data_copy['Fare'].isnull()), 'Fare'].plot.hist(bins = 20)\n",
    "\n",
    "med = np.median(test_data_copy.loc[np.logical_not(test_data_copy['Fare'].isnull()),'Fare'])\n",
    "test_data_copy.loc[test_data_copy['Fare'].isnull(), 'Fare'] = med\n",
    "print(test_data_copy.isnull().sum())\n",
    "\n",
    "\n",
    "plot_train = train_data_copy.copy()\n",
    "plot_test = test_data_copy.copy()\n",
    "small_nonzero = np.min(plot_train.loc[plot_train['Fare'] > 0,'Fare'])\n",
    "\n",
    "lamb_1 = 2\n",
    "plot_train['Fare'] = plot_train['Fare'].apply(lambda x : np.log(x + lamb_1))\n",
    "plot_test['Fare'] = plot_test['Fare'].apply(lambda x : np.log(x + lamb_1))\n",
    "\n",
    "\n",
    "plot_test.loc[np.logical_not(plot_test['Fare'].isnull()), 'Fare'].plot.hist(bins = 10)\n",
    "\n",
    "'''\n",
    "plot_train = train_data_copy.copy()\n",
    "plot_test = test_data_copy.copy()\n",
    "'''"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 225,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0      2.224624\n",
      "1      4.294333\n",
      "2      2.295057\n",
      "3      4.009150\n",
      "4      2.307573\n",
      "5      2.347396\n",
      "6      3.986435\n",
      "7      3.138750\n",
      "8      2.575151\n",
      "9      3.467946\n",
      "10     2.928524\n",
      "11     3.351657\n",
      "12     2.307573\n",
      "13     3.504806\n",
      "14     2.287898\n",
      "15     2.890372\n",
      "16     3.438011\n",
      "17     2.708050\n",
      "18     2.995732\n",
      "19     2.221917\n",
      "20     3.332205\n",
      "21     2.708050\n",
      "22     2.305501\n",
      "23     3.624341\n",
      "24     3.138750\n",
      "25     3.508182\n",
      "26     2.221917\n",
      "27     5.579730\n",
      "28     2.290432\n",
      "29     2.292110\n",
      "         ...   \n",
      "861    2.602690\n",
      "862    3.329673\n",
      "863    4.270397\n",
      "864    2.708050\n",
      "865    2.708050\n",
      "866    2.763693\n",
      "867    3.960733\n",
      "868    2.442347\n",
      "869    2.575151\n",
      "870    2.292110\n",
      "871    3.999195\n",
      "872    1.945910\n",
      "873    2.397895\n",
      "874    3.258097\n",
      "875    2.221917\n",
      "876    2.471973\n",
      "877    2.292110\n",
      "878    2.292110\n",
      "879    4.444512\n",
      "880    3.332205\n",
      "881    2.292110\n",
      "882    2.527064\n",
      "883    2.525729\n",
      "884    2.202765\n",
      "885    3.438011\n",
      "886    2.708050\n",
      "887    3.465736\n",
      "888    3.236716\n",
      "889    3.465736\n",
      "890    2.277267\n",
      "Name: Fare, Length: 891, dtype: float64\n"
     ]
    }
   ],
   "source": [
    "## sort feature of each set to make sure they are in right order\n",
    "\n",
    "plot_train.sort_index(axis=1, inplace=True)\n",
    "plot_test.sort_index(axis = 1, inplace = True)\n",
    "print(plot_train.Fare)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "# Deep Learning NN\n",
    "N-layer NN are used to learn survival of passenger."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 226,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fare\n",
      "Shape of X for train (51, 891)\n",
      "Shape of Y for train (1, 891)\n",
      "Shape of X for test (51, 418)\n",
      "[[ 1.  0.  1. ...,  1.  0.  1.]\n",
      " [ 0.  1.  0. ...,  0.  1.  0.]\n",
      " [ 0.  0.  0. ...,  0.  0.  0.]\n",
      " ..., \n",
      " [ 0.  0.  0. ...,  0.  0.  0.]\n",
      " [ 0.  0.  0. ...,  0.  0.  0.]\n",
      " [ 0.  0.  0. ...,  0.  0.  0.]]\n"
     ]
    }
   ],
   "source": [
    "## create x_train, x_test and y_train\n",
    "train_set_feature = plot_train.drop(['Survived'], axis = 1) # drop label output\n",
    "for col in ['Fare']:\n",
    "    print(col)\n",
    "    plot_test[col] = (plot_test[col]- np.mean(train_set_feature[col]))/np.std(train_set_feature[col])\n",
    "    mean = np.mean(train_set_feature[col])\n",
    "    std = np.std(train_set_feature[col])\n",
    "    train_set_feature[col] = (train_set_feature[col] - mean)/std\n",
    "\n",
    "x_train = train_set_feature.as_matrix().T\n",
    "x_test = plot_test.as_matrix().T\n",
    "\n",
    "y_train = train_data_copy['Survived'].as_matrix().T\n",
    "y_train = y_train.reshape(1, len(y_train))\n",
    "\n",
    "print('Shape of X for train', x_train.shape)\n",
    "print('Shape of Y for train', y_train.shape)\n",
    "print('Shape of X for test', x_test.shape)\n",
    "#print(test_set)\n",
    "print(x_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Create Place Holders (Input, Output)\n",
    "Create method for crate place holder"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 227,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# place holders\n",
    "\n",
    "def create_placeholders(n_x, n_y):\n",
    "    '''\n",
    "    Creates place holders for TensorFlow session (Place holders is a variable that can be update during sessions)\n",
    "    \n",
    "    Arguments:\n",
    "    n_x = Number of input features\n",
    "    n_y = Number of label outputs\n",
    "    \n",
    "    Return:\n",
    "    X = placeholders for input\n",
    "    Y = labels for output\n",
    "    '''\n",
    "    \n",
    "    X = tf.placeholder(tf.float32, shape = (n_x, None), name = 'X')\n",
    "    Y = tf.placeholder(tf.float32, shape = (n_y, None), name = 'Y')\n",
    "    \n",
    "    return X, Y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 228,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X = Tensor(\"X_3:0\", shape=(7, ?), dtype=float32)\n",
      "Y = Tensor(\"Y_3:0\", shape=(1, ?), dtype=float32)\n"
     ]
    }
   ],
   "source": [
    "# Unit test\n",
    "X, Y = create_placeholders(7, 1)\n",
    "print('X =', X)\n",
    "print('Y =', Y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Create Initializing parameters method"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 229,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# create initializing parameters \n",
    "\n",
    "def initialize_parameters_deep(layers_dim, scale_l2):\n",
    "    '''\n",
    "    Create and initialize parameters according to number of hidden neuron in each layers.\n",
    "    \n",
    "    Arguments:\n",
    "    layer_dim = vector of number of hidden neuron in each layer, eg. [n_x, 5, 3, n_y] \n",
    "    \n",
    "    Return:\n",
    "    parameters = dictionary of tensor variables w and b\n",
    "    '''\n",
    "    \n",
    "    parameters = {}\n",
    "        \n",
    "    tf.set_random_seed(1)\n",
    "    num_all_layers = len(layers_dim)\n",
    "\n",
    "    for i in range(num_all_layers-1):\n",
    "        \n",
    "        parameters['W'+str(i+1)] = tf.get_variable('W' + str(i+1), [layers_dim[i+1], layers_dim[i]],\n",
    "                                                   initializer = tf.contrib.layers.xavier_initializer(seed = i))\n",
    "        parameters['b'+str(i+1)] = tf.get_variable('b' + str(i+1), [layers_dim[i+1], 1], initializer = tf.zeros_initializer())\n",
    "        \n",
    "        # sum of all weight to compute regularization\n",
    "        if i == 0:\n",
    "            sum_w = tf.nn.l2_loss(parameters['W'+str(i+1)])\n",
    "        else:\n",
    "            sum_w = sum_w + tf.nn.l2_loss(parameters['W'+str(i+1)])\n",
    "    \n",
    "    sum_w = sum_w*scale_l2\n",
    "    \n",
    "    return parameters, sum_w\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 230,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "W1 <tf.Variable 'W1:0' shape=(7, 5) dtype=float32_ref>\n",
      "b1 <tf.Variable 'b1:0' shape=(7, 1) dtype=float32_ref>\n",
      "W2 <tf.Variable 'W2:0' shape=(3, 7) dtype=float32_ref>\n",
      "b2 <tf.Variable 'b2:0' shape=(3, 1) dtype=float32_ref>\n",
      "sum w Tensor(\"mul:0\", dtype=float32)\n"
     ]
    }
   ],
   "source": [
    "graph = tf.Graph()\n",
    "with graph.as_default(): # reset to default graph every times\n",
    "\n",
    "    scale_l2 = tf.placeholder(tf.float32, name = 'scale_l2')\n",
    "    parameters, sum_w = initialize_parameters_deep([5, 7, 3], scale_l2)\n",
    "    print('W1', parameters['W1'])\n",
    "    print('b1', parameters['b1'])\n",
    "    print('W2', parameters['W2'])\n",
    "    print('b2', parameters['b2'])\n",
    "    print('sum w', sum_w)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "# Forward Propagation Method"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 231,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def forward_propagation(X, parameters, keep_prob, batch_norm = False):\n",
    "    '''\n",
    "    Feed input forward to get estimated output\n",
    "    \n",
    "    Arguments:\n",
    "    X : input\n",
    "    \n",
    "    Return:\n",
    "    ZL : logits output of last layer (not activation output !)\n",
    "    '''\n",
    "    \n",
    "    L = len(parameters)//2 # Number of layers\n",
    "    Atemp = X             # input layer\n",
    "    \n",
    "    if batch_norm == False:\n",
    "        for i in range(L):\n",
    "            Wn = parameters['W' + str(i+1)]\n",
    "            bn = parameters['b' + str(i+1)]\n",
    "            if i < L - 1:\n",
    "                if Wn.shape[0] < 3:\n",
    "                    Ztemp = tf.matmul(Wn, Atemp) + bn\n",
    "                    Atemp = tf.nn.relu(Ztemp)\n",
    "                else:\n",
    "                    Ztemp = tf.nn.dropout(tf.matmul(Wn, Atemp) + bn, keep_prob) # drop out layer for many hidden units\n",
    "                    Atemp = tf.nn.relu(Ztemp)                \n",
    "            else:\n",
    "                Ztemp = tf.matmul(Wn, Atemp) + bn          \n",
    "                Atemp = tf.round(tf.sigmoid(Ztemp))\n",
    "        # return Ztemp because the last process in next method will use this value instead of AL\n",
    "    else:\n",
    "        for i in range(L):\n",
    "            Wn = parameters['W' + str(i+1)]\n",
    "            bn = parameters['b' + str(i+1)]\n",
    "            if i < L - 1:\n",
    "                if Wn.shape[0] < 3:\n",
    "                    Ztemp = tf.matmul(Wn, Atemp) + bn\n",
    "                    Znorm = tf.layers_batch_normalization(Ztemp, axis = 1,\n",
    "                                                          gamma_initializer = tf.contrib.layers.xavier_initializer(seed = i))\n",
    "                    Atemp = tf.nn.tanh(Ztemp)\n",
    "                else:\n",
    "                    Ztemp = tf.nn.dropout(tf.matmul(Wn, Atemp) + bn, keep_prob) # drop out layer for many hidden units\n",
    "                    Atemp = tf.nn.tanh(Ztemp)                \n",
    "            else:\n",
    "                Ztemp = tf.matmul(Wn, Atemp) + bn          \n",
    "                Atemp = tf.round(tf.sigmoid(Ztemp))\n",
    "        # return Ztemp because the last process in next method will use this value instead of AL\n",
    "    \n",
    "    return Ztemp, Atemp\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 232,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tensor(\"mul:0\", dtype=float32)\n"
     ]
    }
   ],
   "source": [
    "graph = tf.Graph()\n",
    "with graph.as_default():\n",
    "    layers_dim = [5, 7, 3, 2]\n",
    "    X,Y = create_placeholders(layers_dim[0], layers_dim[2])\n",
    "    scale_l2 = tf.placeholder(tf.float32, name = 'scale_l2')\n",
    "    keep_prob = tf.placeholder(tf.float32, name = 'keep_prob')\n",
    "    parameters, sum_w = initialize_parameters_deep(layers_dim, scale_l2)\n",
    "    Ztemp, Atemp = forward_propagation(X, parameters, keep_prob)\n",
    "    \n",
    "    print(sum_w)\n",
    "    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Compute Cost Method"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 233,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def compute_cost(ZL, Y, l2_loss):\n",
    "    '''\n",
    "    Compute cost using sigmoid_cross_entropy\n",
    "    \n",
    "    Arguments : \n",
    "    ZL, Y\n",
    "    \n",
    "    Return :\n",
    "    cost\n",
    "    '''\n",
    "    \n",
    "    logits = tf.transpose(ZL)\n",
    "    labels = tf.transpose(Y)\n",
    "    \n",
    "    cost = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits = logits, labels = labels)) + l2_loss\n",
    "    \n",
    "    return cost"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 234,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tensor(\"add_3:0\", dtype=float32)\n"
     ]
    }
   ],
   "source": [
    "graph = tf.Graph()\n",
    "\n",
    "with graph.as_default():\n",
    "    with tf.Session() as sess:\n",
    "        layers_dim = [5, 4, 2]\n",
    "        X, Y = create_placeholders(layers_dim[0], layers_dim[2])\n",
    "        scale_l2 = tf.placeholder(tf.float32, name = 'scale_l2')\n",
    "        keep_prob = tf.placeholder(tf.float32, name = 'keep_prob')\n",
    "        parameters, sum_w = initialize_parameters_deep(layers_dim, scale_l2)\n",
    "        ZL, Atemp = forward_propagation(X, parameters, keep_prob)\n",
    "        cost = compute_cost(ZL, Y, sum_w)\n",
    "\n",
    "    print(cost)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Random Split Train, Dev and Test set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 235,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def random_data(X, Y, ratio):\n",
    "    '''\n",
    "    Random and split train, dev, test set with ratio of given value\n",
    "    \n",
    "    Arguments:\n",
    "    X : input of data\n",
    "    Y : output of data\n",
    "    Ratio : ratio of train, dev and test set\n",
    "    \n",
    "    Return:\n",
    "    Xs : (X_train, X_dev, X_test)\n",
    "    Ys : (Y_train, Y_dev, Y_test)\n",
    "    '''\n",
    "    \n",
    "    m = X.shape[1]\n",
    "    print(m)\n",
    "    randIndex = np.random.permutation(m)\n",
    "    \n",
    "    train_index = randIndex[0:int(ratio[0]*m)]\n",
    "    dev_index = randIndex[int(ratio[0]*m)+1:int((ratio[1]+ratio[0])*m)]\n",
    "    test_index = randIndex[int((ratio[1]+ratio[0])*m)+1:]\n",
    "    \n",
    "    Xs = (X[:, train_index], X[:, dev_index], X[:, test_index])\n",
    "    Ys = (Y[:, train_index], Y[:, dev_index], Y[:, test_index])\n",
    "    \n",
    "    return Xs, Ys"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 236,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "891\n",
      "(51, 0)\n"
     ]
    }
   ],
   "source": [
    "Xs, Ys = random_data(x_train, y_train, [0.9, 0.1, 0.0])\n",
    "print(Xs[2].shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Build N-layer NN !\n",
    "Backpropagation and parameters update are using just 1 line of code, so there is no need to create a method."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 237,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def n_layer_nn(X_train, Y_train, X_dev, Y_dev, X_test, layers_dim_between_n_x_and_n_y, decay = 0.96,\n",
    "               learning_rate = 10**-3, num_epochs = 20000, scale_l2_val = 0.001, keep_prob = 0.9\n",
    "               ,print_cost = True):\n",
    "    \n",
    "    n_x = X_train.shape[0]\n",
    "    n_y = Y_train.shape[0]\n",
    "    \n",
    "    print(Y_train.shape)\n",
    "    \n",
    "    m = Y_train.shape[1]\n",
    "    layers_dim = layers_dim_between_n_x_and_n_y.copy()\n",
    "    layers_dim.insert(0, n_x)\n",
    "    layers_dim.append(n_y)\n",
    "    print('layers_dim', layers_dim)\n",
    "    \n",
    "    ## initialize parameters\n",
    "    graph = tf.Graph()\n",
    "    with graph.as_default(): # set graph as default so this method can be rerun !\n",
    "        X, Y = create_placeholders(n_x, n_y)\n",
    "        \n",
    "        # scale l1 and scale l2\n",
    "        scale_l2_tf = tf.placeholder(tf.float32, name = 'scale_l2')\n",
    "        keep_prob_tf = tf.placeholder(tf.float32, name = 'keep_prob')\n",
    "        \n",
    "        parameters, sum_w = initialize_parameters_deep(layers_dim, scale_l2_tf)\n",
    "        \n",
    "        ZL, AL = forward_propagation(X, parameters, keep_prob_tf)\n",
    "        cost = compute_cost(ZL, Y, sum_w)\n",
    "        \n",
    "        # learning rate decay\n",
    "        global_step = tf.Variable(0, trainable = False)\n",
    "        decay_rate = tf.placeholder(tf.float32, name='decay_rate')\n",
    "        start_learning_rate = learning_rate\n",
    "        learning_rate_decay = tf.train.exponential_decay(start_learning_rate, global_step,\n",
    "                                                        100, decay_rate = decay_rate, staircase = True)\n",
    "        \n",
    "        # create optimizer \n",
    "        optimizer = tf.train.AdamOptimizer(learning_rate = learning_rate_decay).minimize(cost, global_step = global_step)\n",
    "        # initialize all variables \n",
    "        init = tf.global_variables_initializer()\n",
    "        # accuracy evaluation\n",
    "        cor_predict = tf.equal(AL, Y)\n",
    "        accuracy = tf.reduce_mean(tf.cast(cor_predict, tf.float32))\n",
    "    \n",
    "        with tf.Session() as sess:\n",
    "            sess.run(init) # run initialize\n",
    "                \n",
    "            for i in range(num_epochs):\n",
    "                _, cost_val = sess.run([optimizer, cost], feed_dict={X:X_train, Y:Y_train,\n",
    "                                                                     decay_rate:decay, scale_l2_tf:scale_l2_val,\n",
    "                                                                    keep_prob_tf:keep_prob})\n",
    "\n",
    "                if i % 2000 == 0 and print_cost == True:\n",
    "                    print('Cost :', cost_val)\n",
    "\n",
    "            # plot the cost\n",
    "            '''\n",
    "            fig = plt.figure()\n",
    "            plt.plot(np.squeeze(cost_val))\n",
    "            plt.ylabel('cost')\n",
    "            plt.xlabel('iterations')\n",
    "            plt.title(\"Learning rate =\" + str(learning_rate))\n",
    "            plt.show()\n",
    "            '''\n",
    "\n",
    "            # collect parameters\n",
    "            parameters = sess.run(parameters)\n",
    "            Y_test = sess.run(AL, feed_dict={X:X_test, keep_prob_tf:1})\n",
    "\n",
    "            # calculate train accuracy\n",
    "            train_acc = accuracy.eval(feed_dict = {X:X_train, Y:Y_train, keep_prob_tf:1})\n",
    "            dev_acc = accuracy.eval(feed_dict = {X:X_dev, Y:Y_dev, keep_prob_tf:1})\n",
    "\n",
    "            print (\" ---- Train Accuracy ---- :\", train_acc)\n",
    "            print (\" ---- Dev Accuracy ---- :\", dev_acc)\n",
    "\n",
    "            return parameters, train_acc, dev_acc, Y_test\n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 238,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1, 801)\n",
      "layers_dim [51, 22.0, 14.0, 12.0, 1]\n",
      "Cost : 1.1489\n",
      "Cost : 0.471873\n",
      "Cost : 0.463695\n",
      "Cost : 0.467813\n",
      "Cost : 0.461\n",
      "Cost : 0.465016\n",
      "Cost : 0.458385\n",
      "Cost : 0.454543\n",
      "Cost : 0.464928\n",
      "Cost : 0.463446\n",
      " ---- Train Accuracy ---- : 0.858926\n",
      " ---- Dev Accuracy ---- : 0.820225\n"
     ]
    }
   ],
   "source": [
    "parameters, train_acc, dev_acc, Y_test = n_layer_nn(Xs[0], Ys[0], Xs[1], Ys[1], x_test, [22.0, 14.0, 12.0], learning_rate = 0.000516,\n",
    "          scale_l2_val = 0.014041, keep_prob = 0.871057, decay =  0.992619)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 239,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "     Survived\n",
      "0         0.0\n",
      "1         0.0\n",
      "2         1.0\n",
      "3         0.0\n",
      "4         0.0\n",
      "5         0.0\n",
      "6         0.0\n",
      "7         0.0\n",
      "8         1.0\n",
      "9         0.0\n",
      "10        0.0\n",
      "11        1.0\n",
      "12        1.0\n",
      "13        0.0\n",
      "14        1.0\n",
      "15        1.0\n",
      "16        0.0\n",
      "17        0.0\n",
      "18        0.0\n",
      "19        1.0\n",
      "20        0.0\n",
      "21        1.0\n",
      "22        1.0\n",
      "23        0.0\n",
      "24        1.0\n",
      "25        0.0\n",
      "26        1.0\n",
      "27        0.0\n",
      "28        1.0\n",
      "29        0.0\n",
      "..        ...\n",
      "388       0.0\n",
      "389       1.0\n",
      "390       0.0\n",
      "391       1.0\n",
      "392       1.0\n",
      "393       0.0\n",
      "394       0.0\n",
      "395       1.0\n",
      "396       0.0\n",
      "397       1.0\n",
      "398       0.0\n",
      "399       0.0\n",
      "400       1.0\n",
      "401       0.0\n",
      "402       1.0\n",
      "403       0.0\n",
      "404       1.0\n",
      "405       0.0\n",
      "406       0.0\n",
      "407       0.0\n",
      "408       0.0\n",
      "409       1.0\n",
      "410       0.0\n",
      "411       1.0\n",
      "412       0.0\n",
      "413       0.0\n",
      "414       0.0\n",
      "415       0.0\n",
      "416       0.0\n",
      "417       1.0\n",
      "\n",
      "[418 rows x 1 columns]\n"
     ]
    }
   ],
   "source": [
    "Y_test_temp = Y_test.copy()\n",
    "Y_test_temp = np.squeeze(Y_test_temp)\n",
    "\n",
    "dic = {'Survived':Y_test_temp}\n",
    "Y_test_pd = pd.DataFrame(dic)\n",
    "\n",
    "Y_test_pd.to_csv('Output.csv')\n",
    "print(Y_test_pd)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Random Search for Alpha"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def random_search(iteration = 20, print_random = False):\n",
    "    \n",
    "    # random learning rate\n",
    "    alpha = 10**(np.random.rand(iteration)*-3 - 2)\n",
    "    # random number of layer and number of hidden units\n",
    "    num_layer = np.round(np.random.rand(iteration)*6 + 1)\n",
    "    # random decay rate\n",
    "    decay = 1 - 10**(np.random.rand(iteration)*-3 - 0.5)\n",
    "    # random scale_l2\n",
    "    scale_l2 = 10**(np.random.rand(iteration)*-2 - 1)\n",
    "    # keep_prob\n",
    "    keep_prob = 1 - 10**(np.random.rand(iteration)*-1 - 0.349)\n",
    "    \n",
    "    mem = pd.DataFrame(\n",
    "        {'model_number':[],\n",
    "        'learning_rate':[],\n",
    "        'layers_dim':[],\n",
    "        'decay':[],\n",
    "        'scale_l2':[],\n",
    "        'keep_prob':[]})\n",
    "    \n",
    "    for i in range(iteration):\n",
    "        \n",
    "        layers_dim_bet = sorted(list(np.round(np.random.rand(int(num_layer[i]))*20 + 2)), reverse = True)\n",
    "        \n",
    "        if print_random:\n",
    "\n",
    "            print('iter',i)\n",
    "            print('Alpha :', alpha[i])\n",
    "            print('Layer Dim :', layers_dim_bet)\n",
    "            print('decay :', decay[i])\n",
    "            print('scale_l2 :', scale_l2[i])\n",
    "            print('keep_prob:', keep_prob[i])\n",
    "        \n",
    "        memtemp = pd.DataFrame(\n",
    "            {'model_number':'model_' + str(i),\n",
    "            'learning_rate':alpha[i],\n",
    "            'layers_dim': [layers_dim_bet],\n",
    "            'decay':decay[i],\n",
    "            'scale_l2':scale_l2[i],\n",
    "            'keep_prob':keep_prob[i]})\n",
    "        \n",
    "        mem = pd.concat([mem, memtemp], ignore_index = True)\n",
    "        mem2 = mem.set_index('model_number')\n",
    "    \n",
    "    return mem2\n",
    "        \n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[21.0, 12.0, 9.0, 5.0]\n"
     ]
    }
   ],
   "source": [
    "#mem3 = random_search(iteration = 20, print_random = False)\n",
    "print(mem3.loc['model_1','layers_dim'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Cross Validate method"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def create_x_fold(X, Y, num_fold = 10):\n",
    "    \n",
    "    index_permute = np.random.permutation(X.shape[1])\n",
    "    X_temp = X[:]\n",
    "    Y_temp = Y[:]\n",
    "    \n",
    "    X_temp = X_temp[:, index_permute]\n",
    "    Y_temp = Y_temp[0, index_permute]\n",
    "    \n",
    "    X_fold = []\n",
    "    Y_fold = []\n",
    "    \n",
    "    num_each = X_temp.shape[1]//num_fold\n",
    "    \n",
    "    ## create list of X divide to num_fold\n",
    "    for i in range(num_fold):\n",
    "        \n",
    "        if i == num_fold - 1:\n",
    "            X_fold.append(X_temp[:, num_each*i:])\n",
    "            Y_fold.append(Y_temp[num_each*i:])\n",
    "        else:\n",
    "            X_fold.append(X_temp[:, num_each*i:num_each*(i+1)])\n",
    "            Y_fold.append(Y_temp[num_each*i:num_each*(i+1)])\n",
    "        \n",
    "    X_train_out = np.array([])\n",
    "    Y_train_out = np.array([])\n",
    "    \n",
    "    X_out = []\n",
    "    Y_out = []\n",
    "    \n",
    "    \n",
    "    ## create list of tuple for each fold\n",
    "    for i in range(num_fold):\n",
    "        X_dev_out = X_fold[i]\n",
    "        Y_dev_out = Y_fold[i]\n",
    "        \n",
    "        count = 0\n",
    "        \n",
    "        for k in range(num_fold):\n",
    "                \n",
    "            if k != i:\n",
    "                count = count + 1\n",
    "                if count == 1:\n",
    "                    X_train_out = X_fold[k] \n",
    "                    Y_train_out = Y_fold[k]\n",
    "                else:\n",
    "                    X_train_out = np.concatenate((X_train_out, X_fold[k]), axis = 1) \n",
    "                    Y_train_out = np.concatenate((Y_train_out, Y_fold[k]))\n",
    "                    \n",
    "        X_out.append((X_train_out, X_dev_out))\n",
    "        Y_out.append((Y_train_out.reshape(1, len(Y_train_out)), Y_dev_out.reshape(1, len(Y_dev_out))))\n",
    "    \n",
    "    return X_out, Y_out"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0\n",
      "(1, 178)\n",
      "1\n",
      "(1, 178)\n",
      "2\n",
      "(1, 178)\n",
      "3\n",
      "(1, 178)\n",
      "4\n",
      "(1, 179)\n"
     ]
    }
   ],
   "source": [
    "x_train_temp = x_train.copy()\n",
    "y_train_temp = y_train.copy()\n",
    "\n",
    "x_fold, y_fold = create_x_fold(x_train_temp, y_train_temp, num_fold = 5)\n",
    "\n",
    "for i in range(len(x_fold)):\n",
    "    print(i)\n",
    "    print(y_fold[i][1].shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 180,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def cross_validate(X, Y, num_fold = 5, num_random = 20):\n",
    "\n",
    "    random_hyper = random_search(iteration = num_random, print_random = False)\n",
    "    x_fold, y_fold = create_x_fold(X, Y, num_fold = num_fold)\n",
    "    \n",
    "    result_fold = pd.DataFrame({})\n",
    "    \n",
    "    for i in range(num_random): # loop through each model\n",
    "        \n",
    "        # retrive decay, keep_prob, learning_rate, scale_l2 and layers_dim\n",
    "        layers_dim = random_hyper.loc['model_'+str(i), 'layers_dim']\n",
    "        decay = random_hyper.loc['model_'+str(i), 'decay']\n",
    "        keep_prob = random_hyper.loc['model_'+str(i), 'keep_prob']\n",
    "\n",
    "        learning_rate = random_hyper.loc['model_'+str(i), 'learning_rate']\n",
    "        scale_l2 = random_hyper.loc['model_'+str(i), 'scale_l2']\n",
    "\n",
    "        dic = {}\n",
    "        result_each_best = pd.DataFrame({})\n",
    "            \n",
    "        for k in range(num_fold): # loop through each fold\n",
    "\n",
    "            print('*** Fold Number ***' + str(k+1) + ' ---')\n",
    "            print('*** Model Number ***' + str(i+1) + ' ---')\n",
    "\n",
    "            X_train = x_fold[k][0]\n",
    "            X_dev = x_fold[k][1]\n",
    "\n",
    "            Y_train = y_fold[k][0]\n",
    "            Y_dev = y_fold[k][1]      \n",
    "\n",
    "            _, train_acc, dev_acc, Y_test = n_layer_nn(X_train, Y_train, X_dev, Y_dev, X_train, layers_dim, learning_rate = learning_rate,\n",
    "                       scale_l2_val = scale_l2, keep_prob =keep_prob, decay = decay )\n",
    "\n",
    "            dic['model_'+str(i)+'_dev_acc'] = [dev_acc]\n",
    "            dic['model_'+str(i)+'_train_acc'] = [train_acc]\n",
    "\n",
    "            result_each_best = pd.concat([result_each_best, pd.DataFrame(dic)], ignore_index = True)\n",
    "\n",
    "        mean_dev = np.mean(result_each_best['model_'+str(i)+'_dev_acc'])\n",
    "        mean_train = np.mean(result_each_best['model_'+str(i)+'_train_acc'])\n",
    "\n",
    "        dic['model_'+str(i)+'_dev_acc'] = [mean_dev]\n",
    "        dic['model_'+str(i)+'_train_acc'] = [mean_train]\n",
    "        result_each_best = pd.concat([result_each_best, pd.DataFrame(dic)], ignore_index = True)\n",
    "\n",
    "        std_dev = np.std(result_each_best['model_'+str(i)+'_dev_acc'])\n",
    "        std_train = np.std(result_each_best['model_'+str(i)+'_train_acc'])       \n",
    "\n",
    "        dic['model_'+str(i)+'_dev_acc'] = [std_dev]\n",
    "        dic['model_'+str(i)+'_train_acc'] = [std_train]\n",
    "        result_each_best = pd.concat([result_each_best, pd.DataFrame(dic)], ignore_index = True)\n",
    "\n",
    "        result_fold = pd.concat([result_fold, result_each_best], axis = 1, ignore_index = False)\n",
    "    \n",
    "    ## append row index\n",
    "    dic = {}\n",
    "    fold = []\n",
    "    for i in range(num_fold):\n",
    "        fold.append('fold_'+str(i))\n",
    "        \n",
    "    fold.append('mean')\n",
    "    fold.append('std')\n",
    "    dic['fold'] = fold\n",
    "    \n",
    "    result_fold = pd.concat([pd.DataFrame(dic), result_fold], axis = 1, ignore_index = False)\n",
    "    result_fold_output = result_fold.set_index('fold')\n",
    "\n",
    "    print(result_fold_output)\n",
    "    \n",
    "    return result_fold_output, random_hyper"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 181,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "*** Fold Number ***1 ---\n",
      "*** Model Number ***1 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 6.0, 6.0, 5.0, 1]\n",
      "Cost : 1.71049\n",
      "Cost : 0.667737\n",
      "Cost : 0.667742\n",
      "Cost : 0.667742\n",
      "Cost : 0.667745\n",
      "Cost : 0.66775\n",
      "Cost : 0.667747\n",
      "Cost : 0.667737\n",
      "Cost : 0.667745\n",
      "Cost : 0.667741\n",
      " ---- Train Accuracy ---- : 0.612219\n",
      " ---- Dev Accuracy ---- : 0.651685\n",
      "*** Fold Number ***2 ---\n",
      "*** Model Number ***1 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 6.0, 6.0, 5.0, 1]\n",
      "Cost : 1.70835\n",
      "Cost : 0.668303\n",
      "Cost : 0.668309\n",
      "Cost : 0.668301\n",
      "Cost : 0.668321\n",
      "Cost : 0.668314\n",
      "Cost : 0.668309\n",
      "Cost : 0.668295\n",
      "Cost : 0.668311\n",
      "Cost : 0.668316\n",
      " ---- Train Accuracy ---- : 0.610973\n",
      " ---- Dev Accuracy ---- : 0.662921\n",
      "*** Fold Number ***3 ---\n",
      "*** Model Number ***1 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 6.0, 6.0, 5.0, 1]\n",
      "Cost : 1.70975\n",
      "Cost : 0.665398\n",
      "Cost : 0.665413\n",
      "Cost : 0.665399\n",
      "Cost : 0.665426\n",
      "Cost : 0.665415\n",
      "Cost : 0.665413\n",
      "Cost : 0.665403\n",
      "Cost : 0.665414\n",
      "Cost : 0.665425\n",
      " ---- Train Accuracy ---- : 0.617207\n",
      " ---- Dev Accuracy ---- : 0.606742\n",
      "*** Fold Number ***4 ---\n",
      "*** Model Number ***1 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 6.0, 6.0, 5.0, 1]\n",
      "Cost : 1.70867\n",
      "Cost : 0.663576\n",
      "Cost : 0.663599\n",
      "Cost : 0.663577\n",
      "Cost : 0.663609\n",
      "Cost : 0.663596\n",
      "Cost : 0.663604\n",
      "Cost : 0.663584\n",
      "Cost : 0.663598\n",
      "Cost : 0.663598\n",
      " ---- Train Accuracy ---- : 0.620948\n",
      " ---- Dev Accuracy ---- : 0.573034\n",
      "*** Fold Number ***5 ---\n",
      "*** Model Number ***1 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 6.0, 6.0, 5.0, 1]\n",
      "Cost : 1.70784\n",
      "Cost : 0.665397\n",
      "Cost : 0.665415\n",
      "Cost : 0.665392\n",
      "Cost : 0.665415\n",
      "Cost : 0.665418\n",
      "Cost : 0.66542\n",
      "Cost : 0.665397\n",
      "Cost : 0.665412\n",
      "Cost : 0.665431\n",
      " ---- Train Accuracy ---- : 0.617207\n",
      " ---- Dev Accuracy ---- : 0.606742\n",
      "*** Fold Number ***6 ---\n",
      "*** Model Number ***1 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 6.0, 6.0, 5.0, 1]\n",
      "Cost : 1.70581\n",
      "Cost : 0.667158\n",
      "Cost : 0.66717\n",
      "Cost : 0.667159\n",
      "Cost : 0.667174\n",
      "Cost : 0.667173\n",
      "Cost : 0.667168\n",
      "Cost : 0.667166\n",
      "Cost : 0.667171\n",
      "Cost : 0.66718\n",
      " ---- Train Accuracy ---- : 0.613466\n",
      " ---- Dev Accuracy ---- : 0.640449\n",
      "*** Fold Number ***7 ---\n",
      "*** Model Number ***1 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 6.0, 6.0, 5.0, 1]\n",
      "Cost : 1.703\n",
      "Cost : 0.664197\n",
      "Cost : 0.664211\n",
      "Cost : 0.664204\n",
      "Cost : 0.664219\n",
      "Cost : 0.664209\n",
      "Cost : 0.664205\n",
      "Cost : 0.664206\n",
      "Cost : 0.664209\n",
      "Cost : 0.664221\n",
      " ---- Train Accuracy ---- : 0.619701\n",
      " ---- Dev Accuracy ---- : 0.58427\n",
      "*** Fold Number ***8 ---\n",
      "*** Model Number ***1 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 6.0, 6.0, 5.0, 1]\n",
      "Cost : 1.70204\n",
      "Cost : 0.665401\n",
      "Cost : 0.665422\n",
      "Cost : 0.66541\n",
      "Cost : 0.665417\n",
      "Cost : 0.665415\n",
      "Cost : 0.665418\n",
      "Cost : 0.665415\n",
      "Cost : 0.665409\n",
      "Cost : 0.665418\n",
      " ---- Train Accuracy ---- : 0.617207\n",
      " ---- Dev Accuracy ---- : 0.606742\n",
      "*** Fold Number ***9 ---\n",
      "*** Model Number ***1 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 6.0, 6.0, 5.0, 1]\n",
      "Cost : 1.70296\n",
      "Cost : 0.665995\n",
      "Cost : 0.666001\n",
      "Cost : 0.666004\n",
      "Cost : 0.666008\n",
      "Cost : 0.666006\n",
      "Cost : 0.666019\n",
      "Cost : 0.666001\n",
      "Cost : 0.666001\n",
      "Cost : 0.666012\n",
      " ---- Train Accuracy ---- : 0.61596\n",
      " ---- Dev Accuracy ---- : 0.617977\n",
      "*** Fold Number ***10 ---\n",
      "*** Model Number ***1 ---\n",
      "(1, 801)\n",
      "layers_dim [51, 6.0, 6.0, 5.0, 1]\n",
      "Cost : 1.7127\n",
      "Cost : 0.665643\n",
      "Cost : 0.665646\n",
      "Cost : 0.665647\n",
      "Cost : 0.665634\n",
      "Cost : 0.665637\n",
      "Cost : 0.665641\n",
      "Cost : 0.66565\n",
      "Cost : 0.665626\n",
      "Cost : 0.665654\n",
      " ---- Train Accuracy ---- : 0.616729\n",
      " ---- Dev Accuracy ---- : 0.611111\n",
      "*** Fold Number ***1 ---\n",
      "*** Model Number ***2 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 22.0, 18.0, 7.0, 1]\n",
      "Cost : 2.23943\n",
      "Cost : 0.877116\n",
      "Cost : 0.666039\n",
      "Cost : 0.634835\n",
      "Cost : 0.617458\n",
      "Cost : 0.617941\n",
      "Cost : 0.610001\n",
      "Cost : 0.607984\n",
      "Cost : 0.607333\n",
      "Cost : 0.630628\n",
      " ---- Train Accuracy ---- : 0.809227\n",
      " ---- Dev Accuracy ---- : 0.797753\n",
      "*** Fold Number ***2 ---\n",
      "*** Model Number ***2 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 22.0, 18.0, 7.0, 1]\n",
      "Cost : 2.23646\n",
      "Cost : 0.871131\n",
      "Cost : 0.660522\n",
      "Cost : 0.634718\n",
      "Cost : 0.609727\n",
      "Cost : 0.61181\n",
      "Cost : 0.599994\n",
      "Cost : 0.607954\n",
      "Cost : 0.598564\n",
      "Cost : 0.625324\n",
      " ---- Train Accuracy ---- : 0.812968\n",
      " ---- Dev Accuracy ---- : 0.775281\n",
      "*** Fold Number ***3 ---\n",
      "*** Model Number ***2 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 22.0, 18.0, 7.0, 1]\n",
      "Cost : 2.24295\n",
      "Cost : 0.870251\n",
      "Cost : 0.665889\n",
      "Cost : 0.636489\n",
      "Cost : 0.612328\n",
      "Cost : 0.612432\n",
      "Cost : 0.603275\n",
      "Cost : 0.61001\n",
      "Cost : 0.604702\n",
      "Cost : 0.629098\n",
      " ---- Train Accuracy ---- : 0.814214\n",
      " ---- Dev Accuracy ---- : 0.820225\n",
      "*** Fold Number ***4 ---\n",
      "*** Model Number ***2 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 22.0, 18.0, 7.0, 1]\n",
      "Cost : 2.24298\n",
      "Cost : 0.867229\n",
      "Cost : 0.660973\n",
      "Cost : 0.630891\n",
      "Cost : 0.612779\n",
      "Cost : 0.604106\n",
      "Cost : 0.599013\n",
      "Cost : 0.598266\n",
      "Cost : 0.59582\n",
      "Cost : 0.623446\n",
      " ---- Train Accuracy ---- : 0.819202\n",
      " ---- Dev Accuracy ---- : 0.775281\n",
      "*** Fold Number ***5 ---\n",
      "*** Model Number ***2 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 22.0, 18.0, 7.0, 1]\n",
      "Cost : 2.23847\n",
      "Cost : 0.871044\n",
      "Cost : 0.674293\n",
      "Cost : 0.634968\n",
      "Cost : 0.618205\n",
      "Cost : 0.611585\n",
      "Cost : 0.616718\n",
      "Cost : 0.604313\n",
      "Cost : 0.609663\n",
      "Cost : 0.634342\n",
      " ---- Train Accuracy ---- : 0.814214\n",
      " ---- Dev Accuracy ---- : 0.842697\n",
      "*** Fold Number ***6 ---\n",
      "*** Model Number ***2 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 22.0, 18.0, 7.0, 1]\n",
      "Cost : 2.24302\n",
      "Cost : 0.868561\n",
      "Cost : 0.676327\n",
      "Cost : 0.639413\n",
      "Cost : 0.616513\n",
      "Cost : 0.617964\n",
      "Cost : 0.619327\n",
      "Cost : 0.604862\n",
      "Cost : 0.608591\n",
      "Cost : 0.640235\n",
      " ---- Train Accuracy ---- : 0.80798\n",
      " ---- Dev Accuracy ---- : 0.842697\n",
      "*** Fold Number ***7 ---\n",
      "*** Model Number ***2 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 22.0, 18.0, 7.0, 1]\n",
      "Cost : 2.23999\n",
      "Cost : 0.860446\n",
      "Cost : 0.672406\n",
      "Cost : 0.632369\n",
      "Cost : 0.605325\n",
      "Cost : 0.602316\n",
      "Cost : 0.612214\n",
      "Cost : 0.602129\n",
      "Cost : 0.605782\n",
      "Cost : 0.631133\n",
      " ---- Train Accuracy ---- : 0.817955\n",
      " ---- Dev Accuracy ---- : 0.786517\n",
      "*** Fold Number ***8 ---\n",
      "*** Model Number ***2 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 22.0, 18.0, 7.0, 1]\n",
      "Cost : 2.23968\n",
      "Cost : 0.863863\n",
      "Cost : 0.672424\n",
      "Cost : 0.633992\n",
      "Cost : 0.607529\n",
      "Cost : 0.599214\n",
      "Cost : 0.611563\n",
      "Cost : 0.602009\n",
      "Cost : 0.613533\n",
      "Cost : 0.633164\n",
      " ---- Train Accuracy ---- : 0.815461\n",
      " ---- Dev Accuracy ---- : 0.797753\n",
      "*** Fold Number ***9 ---\n",
      "*** Model Number ***2 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 22.0, 18.0, 7.0, 1]\n",
      "Cost : 2.23942\n",
      "Cost : 0.864373\n",
      "Cost : 0.677937\n",
      "Cost : 0.632869\n",
      "Cost : 0.609205\n",
      "Cost : 0.600479\n",
      "Cost : 0.615166\n",
      "Cost : 0.609251\n",
      "Cost : 0.622286\n",
      "Cost : 0.634669\n",
      " ---- Train Accuracy ---- : 0.816708\n",
      " ---- Dev Accuracy ---- : 0.820225\n",
      "*** Fold Number ***10 ---\n",
      "*** Model Number ***2 ---\n",
      "(1, 801)\n",
      "layers_dim [51, 22.0, 18.0, 7.0, 1]\n",
      "Cost : 2.23307\n",
      "Cost : 0.878591\n",
      "Cost : 0.669848\n",
      "Cost : 0.627451\n",
      "Cost : 0.619897\n",
      "Cost : 0.613953\n",
      "Cost : 0.624605\n",
      "Cost : 0.608188\n",
      "Cost : 0.610148\n",
      "Cost : 0.626008\n",
      " ---- Train Accuracy ---- : 0.818976\n",
      " ---- Dev Accuracy ---- : 0.788889\n",
      "*** Fold Number ***1 ---\n",
      "*** Model Number ***3 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 16.0, 15.0, 12.0, 12.0, 7.0, 7.0, 1]\n",
      "Cost : 1.00135\n",
      "Cost : 0.485777\n",
      "Cost : 0.456902\n",
      "Cost : 0.452528\n",
      "Cost : 0.433799\n",
      "Cost : 0.447082\n",
      "Cost : 0.444359\n",
      "Cost : 0.450158\n",
      "Cost : 0.442156\n",
      "Cost : 0.466534\n",
      " ---- Train Accuracy ---- : 0.872818\n",
      " ---- Dev Accuracy ---- : 0.876405\n",
      "*** Fold Number ***2 ---\n",
      "*** Model Number ***3 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 16.0, 15.0, 12.0, 12.0, 7.0, 7.0, 1]\n",
      "Cost : 0.999054\n",
      "Cost : 0.46735\n",
      "Cost : 0.443046\n",
      "Cost : 0.457711\n",
      "Cost : 0.422201\n",
      "Cost : 0.440984\n",
      "Cost : 0.423002\n",
      "Cost : 0.432779\n",
      "Cost : 0.433812\n",
      "Cost : 0.442724\n",
      " ---- Train Accuracy ---- : 0.880299\n",
      " ---- Dev Accuracy ---- : 0.853933\n",
      "*** Fold Number ***3 ---\n",
      "*** Model Number ***3 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 16.0, 15.0, 12.0, 12.0, 7.0, 7.0, 1]\n",
      "Cost : 0.99832\n",
      "Cost : 0.473835\n",
      "Cost : 0.43176\n",
      "Cost : 0.441871\n",
      "Cost : 0.413645\n",
      "Cost : 0.427821\n",
      "Cost : 0.430051\n",
      "Cost : 0.435248\n",
      "Cost : 0.428482\n",
      "Cost : 0.437248\n",
      " ---- Train Accuracy ---- : 0.876559\n",
      " ---- Dev Accuracy ---- : 0.808989\n",
      "*** Fold Number ***4 ---\n",
      "*** Model Number ***3 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 16.0, 15.0, 12.0, 12.0, 7.0, 7.0, 1]\n",
      "Cost : 0.998628\n",
      "Cost : 0.477341\n",
      "Cost : 0.436996\n",
      "Cost : 0.453153\n",
      "Cost : 0.43738\n",
      "Cost : 0.433684\n",
      "Cost : 0.42727\n",
      "Cost : 0.428125\n",
      "Cost : 0.430122\n",
      "Cost : 0.4324\n",
      " ---- Train Accuracy ---- : 0.885287\n",
      " ---- Dev Accuracy ---- : 0.764045\n",
      "*** Fold Number ***5 ---\n",
      "*** Model Number ***3 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 16.0, 15.0, 12.0, 12.0, 7.0, 7.0, 1]\n",
      "Cost : 0.995823\n",
      "Cost : 0.482806\n",
      "Cost : 0.451745\n",
      "Cost : 0.454136\n",
      "Cost : 0.449762\n",
      "Cost : 0.444545\n",
      "Cost : 0.441452\n",
      "Cost : 0.439365\n",
      "Cost : 0.438896\n",
      "Cost : 0.441655\n",
      " ---- Train Accuracy ---- : 0.882793\n",
      " ---- Dev Accuracy ---- : 0.820225\n",
      "*** Fold Number ***6 ---\n",
      "*** Model Number ***3 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 16.0, 15.0, 12.0, 12.0, 7.0, 7.0, 1]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cost : 0.993901\n",
      "Cost : 0.491071\n",
      "Cost : 0.455143\n",
      "Cost : 0.45834\n",
      "Cost : 0.455437\n",
      "Cost : 0.447673\n",
      "Cost : 0.450625\n",
      "Cost : 0.448701\n",
      "Cost : 0.444057\n",
      "Cost : 0.451912\n",
      " ---- Train Accuracy ---- : 0.877805\n",
      " ---- Dev Accuracy ---- : 0.853933\n",
      "*** Fold Number ***7 ---\n",
      "*** Model Number ***3 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 16.0, 15.0, 12.0, 12.0, 7.0, 7.0, 1]\n",
      "Cost : 0.991886\n",
      "Cost : 0.485173\n",
      "Cost : 0.440332\n",
      "Cost : 0.440882\n",
      "Cost : 0.44412\n",
      "Cost : 0.4283\n",
      "Cost : 0.440345\n",
      "Cost : 0.438103\n",
      "Cost : 0.429137\n",
      "Cost : 0.430337\n",
      " ---- Train Accuracy ---- : 0.885287\n",
      " ---- Dev Accuracy ---- : 0.808989\n",
      "*** Fold Number ***8 ---\n",
      "*** Model Number ***3 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 16.0, 15.0, 12.0, 12.0, 7.0, 7.0, 1]\n",
      "Cost : 0.994376\n",
      "Cost : 0.480772\n",
      "Cost : 0.435816\n",
      "Cost : 0.434825\n",
      "Cost : 0.438025\n",
      "Cost : 0.43103\n",
      "Cost : 0.433879\n",
      "Cost : 0.435211\n",
      "Cost : 0.428917\n",
      "Cost : 0.437827\n",
      " ---- Train Accuracy ---- : 0.881546\n",
      " ---- Dev Accuracy ---- : 0.808989\n",
      "*** Fold Number ***9 ---\n",
      "*** Model Number ***3 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 16.0, 15.0, 12.0, 12.0, 7.0, 7.0, 1]\n",
      "Cost : 0.991586\n",
      "Cost : 0.490687\n",
      "Cost : 0.461491\n",
      "Cost : 0.442761\n",
      "Cost : 0.460929\n",
      "Cost : 0.438208\n",
      "Cost : 0.448807\n",
      "Cost : 0.434208\n",
      "Cost : 0.436137\n",
      "Cost : 0.435776\n",
      " ---- Train Accuracy ---- : 0.875312\n",
      " ---- Dev Accuracy ---- : 0.820225\n",
      "*** Fold Number ***10 ---\n",
      "*** Model Number ***3 ---\n",
      "(1, 801)\n",
      "layers_dim [51, 20.0, 16.0, 15.0, 12.0, 12.0, 7.0, 7.0, 1]\n",
      "Cost : 0.998039\n",
      "Cost : 0.466037\n",
      "Cost : 0.444163\n",
      "Cost : 0.459137\n",
      "Cost : 0.4357\n",
      "Cost : 0.430276\n",
      "Cost : 0.439561\n",
      "Cost : 0.422323\n",
      "Cost : 0.429493\n",
      "Cost : 0.422445\n",
      " ---- Train Accuracy ---- : 0.882647\n",
      " ---- Dev Accuracy ---- : 0.811111\n",
      "*** Fold Number ***1 ---\n",
      "*** Model Number ***4 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 20.0, 20.0, 18.0, 17.0, 15.0, 3.0, 1]\n",
      "Cost : 3.76983\n",
      "Cost : 0.787502\n",
      "Cost : 0.673218\n",
      "Cost : 0.668752\n",
      "Cost : 0.667848\n",
      "Cost : 0.667744\n",
      "Cost : 0.667745\n",
      "Cost : 0.667745\n",
      "Cost : 0.667745\n",
      "Cost : 0.667745\n",
      " ---- Train Accuracy ---- : 0.612219\n",
      " ---- Dev Accuracy ---- : 0.651685\n",
      "*** Fold Number ***2 ---\n",
      "*** Model Number ***4 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 20.0, 20.0, 18.0, 17.0, 15.0, 3.0, 1]\n",
      "Cost : 3.76983\n",
      "Cost : 0.787885\n",
      "Cost : 0.673766\n",
      "Cost : 0.669316\n",
      "Cost : 0.668417\n",
      "Cost : 0.668313\n",
      "Cost : 0.668312\n",
      "Cost : 0.668312\n",
      "Cost : 0.668312\n",
      "Cost : 0.668312\n",
      " ---- Train Accuracy ---- : 0.610973\n",
      " ---- Dev Accuracy ---- : 0.662921\n",
      "*** Fold Number ***3 ---\n",
      "*** Model Number ***4 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 20.0, 20.0, 18.0, 17.0, 15.0, 3.0, 1]\n",
      "Cost : 3.77081\n",
      "Cost : 0.786129\n",
      "Cost : 0.671\n",
      "Cost : 0.66644\n",
      "Cost : 0.665523\n",
      "Cost : 0.665416\n",
      "Cost : 0.665415\n",
      "Cost : 0.665415\n",
      "Cost : 0.665415\n",
      "Cost : 0.665415\n",
      " ---- Train Accuracy ---- : 0.617207\n",
      " ---- Dev Accuracy ---- : 0.606742\n",
      "*** Fold Number ***4 ---\n",
      "*** Model Number ***4 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 20.0, 20.0, 18.0, 17.0, 15.0, 3.0, 1]\n",
      "Cost : 3.77092\n",
      "Cost : 0.784826\n",
      "Cost : 0.669231\n",
      "Cost : 0.664613\n",
      "Cost : 0.663704\n",
      "Cost : 0.663599\n",
      "Cost : 0.663598\n",
      "Cost : 0.663598\n",
      "Cost : 0.663598\n",
      "Cost : 0.663598\n",
      " ---- Train Accuracy ---- : 0.620948\n",
      " ---- Dev Accuracy ---- : 0.573034\n",
      "*** Fold Number ***5 ---\n",
      "*** Model Number ***4 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 20.0, 20.0, 18.0, 17.0, 15.0, 3.0, 1]\n",
      "Cost : 3.77131\n",
      "Cost : 0.785973\n",
      "Cost : 0.670971\n",
      "Cost : 0.666427\n",
      "Cost : 0.665521\n",
      "Cost : 0.665416\n",
      "Cost : 0.665415\n",
      "Cost : 0.665415\n",
      "Cost : 0.665415\n",
      "Cost : 0.665415\n",
      " ---- Train Accuracy ---- : 0.617207\n",
      " ---- Dev Accuracy ---- : 0.606742\n",
      "*** Fold Number ***6 ---\n",
      "*** Model Number ***4 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 20.0, 20.0, 18.0, 17.0, 15.0, 3.0, 1]\n",
      "Cost : 3.77169\n",
      "Cost : 0.787252\n",
      "Cost : 0.672688\n",
      "Cost : 0.668191\n",
      "Cost : 0.667281\n",
      "Cost : 0.667175\n",
      "Cost : 0.667173\n",
      "Cost : 0.667173\n",
      "Cost : 0.667173\n",
      "Cost : 0.667173\n",
      " ---- Train Accuracy ---- : 0.613466\n",
      " ---- Dev Accuracy ---- : 0.640449\n",
      "*** Fold Number ***7 ---\n",
      "*** Model Number ***4 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 20.0, 20.0, 18.0, 17.0, 15.0, 3.0, 1]\n",
      "Cost : 3.77152\n",
      "Cost : 0.785206\n",
      "Cost : 0.669818\n",
      "Cost : 0.665226\n",
      "Cost : 0.664318\n",
      "Cost : 0.664211\n",
      "Cost : 0.66421\n",
      "Cost : 0.66421\n",
      "Cost : 0.66421\n",
      "Cost : 0.66421\n",
      " ---- Train Accuracy ---- : 0.619701\n",
      " ---- Dev Accuracy ---- : 0.58427\n",
      "*** Fold Number ***8 ---\n",
      "*** Model Number ***4 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 20.0, 20.0, 18.0, 17.0, 15.0, 3.0, 1]\n",
      "Cost : 3.77139\n",
      "Cost : 0.785877\n",
      "Cost : 0.670955\n",
      "Cost : 0.666422\n",
      "Cost : 0.66552\n",
      "Cost : 0.665416\n",
      "Cost : 0.665415\n",
      "Cost : 0.665415\n",
      "Cost : 0.665415\n",
      "Cost : 0.665415\n",
      " ---- Train Accuracy ---- : 0.617207\n",
      " ---- Dev Accuracy ---- : 0.606742\n",
      "*** Fold Number ***9 ---\n",
      "*** Model Number ***4 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 20.0, 20.0, 18.0, 17.0, 15.0, 3.0, 1]\n",
      "Cost : 3.77275\n",
      "Cost : 0.786369\n",
      "Cost : 0.671539\n",
      "Cost : 0.667017\n",
      "Cost : 0.666113\n",
      "Cost : 0.666008\n",
      "Cost : 0.666007\n",
      "Cost : 0.666007\n",
      "Cost : 0.666007\n",
      "Cost : 0.666007\n",
      " ---- Train Accuracy ---- : 0.61596\n",
      " ---- Dev Accuracy ---- : 0.617977\n",
      "*** Fold Number ***10 ---\n",
      "*** Model Number ***4 ---\n",
      "(1, 801)\n",
      "layers_dim [51, 20.0, 20.0, 20.0, 18.0, 17.0, 15.0, 3.0, 1]\n",
      "Cost : 3.77081\n",
      "Cost : 0.78618\n",
      "Cost : 0.6712\n",
      "Cost : 0.666658\n",
      "Cost : 0.665751\n",
      "Cost : 0.665643\n",
      "Cost : 0.665642\n",
      "Cost : 0.665642\n",
      "Cost : 0.665642\n",
      "Cost : 0.665642\n",
      " ---- Train Accuracy ---- : 0.616729\n",
      " ---- Dev Accuracy ---- : 0.611111\n",
      "*** Fold Number ***1 ---\n",
      "*** Model Number ***5 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 13.0, 1]\n",
      "Cost : 1.12295\n",
      "Cost : 0.972472\n",
      "Cost : 0.899914\n",
      "Cost : 0.856447\n",
      "Cost : 0.830126\n",
      "Cost : 0.815805\n",
      "Cost : 0.803312\n",
      "Cost : 0.797993\n",
      "Cost : 0.794981\n",
      "Cost : 0.791479\n",
      " ---- Train Accuracy ---- : 0.763092\n",
      " ---- Dev Accuracy ---- : 0.842697\n",
      "*** Fold Number ***2 ---\n",
      "*** Model Number ***5 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 13.0, 1]\n",
      "Cost : 1.12057\n",
      "Cost : 0.96858\n",
      "Cost : 0.895354\n",
      "Cost : 0.852421\n",
      "Cost : 0.824656\n",
      "Cost : 0.811402\n",
      "Cost : 0.796968\n",
      "Cost : 0.793296\n",
      "Cost : 0.789428\n",
      "Cost : 0.78517\n",
      " ---- Train Accuracy ---- : 0.765586\n",
      " ---- Dev Accuracy ---- : 0.797753\n",
      "*** Fold Number ***3 ---\n",
      "*** Model Number ***5 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 13.0, 1]\n",
      "Cost : 1.12001\n",
      "Cost : 0.968109\n",
      "Cost : 0.895425\n",
      "Cost : 0.851403\n",
      "Cost : 0.825496\n",
      "Cost : 0.811708\n",
      "Cost : 0.798069\n",
      "Cost : 0.793531\n",
      "Cost : 0.789459\n",
      "Cost : 0.786263\n",
      " ---- Train Accuracy ---- : 0.76808\n",
      " ---- Dev Accuracy ---- : 0.752809\n",
      "*** Fold Number ***4 ---\n",
      "*** Model Number ***5 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 13.0, 1]\n",
      "Cost : 1.11392\n",
      "Cost : 0.961538\n",
      "Cost : 0.886243\n",
      "Cost : 0.844197\n",
      "Cost : 0.817931\n",
      "Cost : 0.803201\n",
      "Cost : 0.790895\n",
      "Cost : 0.784286\n",
      "Cost : 0.77895\n",
      "Cost : 0.778622\n",
      " ---- Train Accuracy ---- : 0.778055\n",
      " ---- Dev Accuracy ---- : 0.651685\n",
      "*** Fold Number ***5 ---\n",
      "*** Model Number ***5 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 13.0, 1]\n",
      "Cost : 1.11367\n",
      "Cost : 0.964103\n",
      "Cost : 0.889355\n",
      "Cost : 0.849695\n",
      "Cost : 0.823979\n",
      "Cost : 0.808144\n",
      "Cost : 0.796892\n",
      "Cost : 0.790812\n",
      "Cost : 0.782978\n",
      "Cost : 0.783355\n",
      " ---- Train Accuracy ---- : 0.763092\n",
      " ---- Dev Accuracy ---- : 0.775281\n",
      "*** Fold Number ***6 ---\n",
      "*** Model Number ***5 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 13.0, 1]\n",
      "Cost : 1.124\n",
      "Cost : 0.97318\n",
      "Cost : 0.899595\n",
      "Cost : 0.858369\n",
      "Cost : 0.833128\n",
      "Cost : 0.817833\n",
      "Cost : 0.805993\n",
      "Cost : 0.799849\n",
      "Cost : 0.791876\n",
      "Cost : 0.792702\n",
      " ---- Train Accuracy ---- : 0.764339\n",
      " ---- Dev Accuracy ---- : 0.831461\n",
      "*** Fold Number ***7 ---\n",
      "*** Model Number ***5 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 13.0, 1]\n",
      "Cost : 1.11916\n",
      "Cost : 0.966285\n",
      "Cost : 0.892275\n",
      "Cost : 0.850031\n",
      "Cost : 0.823599\n",
      "Cost : 0.808972\n",
      "Cost : 0.796453\n",
      "Cost : 0.789716\n",
      "Cost : 0.783226\n",
      "Cost : 0.784143\n",
      " ---- Train Accuracy ---- : 0.76808\n",
      " ---- Dev Accuracy ---- : 0.741573\n",
      "*** Fold Number ***8 ---\n",
      "*** Model Number ***5 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 13.0, 1]\n",
      "Cost : 1.12236\n",
      "Cost : 0.968268\n",
      "Cost : 0.8935\n",
      "Cost : 0.85319\n",
      "Cost : 0.825577\n",
      "Cost : 0.810682\n",
      "Cost : 0.798142\n",
      "Cost : 0.792054\n",
      "Cost : 0.784072\n",
      "Cost : 0.785797\n",
      " ---- Train Accuracy ---- : 0.765586\n",
      " ---- Dev Accuracy ---- : 0.752809\n",
      "*** Fold Number ***9 ---\n",
      "*** Model Number ***5 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 13.0, 1]\n",
      "Cost : 1.11707\n",
      "Cost : 0.965502\n",
      "Cost : 0.891006\n",
      "Cost : 0.850009\n",
      "Cost : 0.822802\n",
      "Cost : 0.810036\n",
      "Cost : 0.796991\n",
      "Cost : 0.791092\n",
      "Cost : 0.780725\n",
      "Cost : 0.783604\n",
      " ---- Train Accuracy ---- : 0.763092\n",
      " ---- Dev Accuracy ---- : 0.752809\n",
      "*** Fold Number ***10 ---\n",
      "*** Model Number ***5 ---\n",
      "(1, 801)\n",
      "layers_dim [51, 13.0, 1]\n",
      "Cost : 1.12763\n",
      "Cost : 0.97425\n",
      "Cost : 0.895018\n",
      "Cost : 0.849394\n",
      "Cost : 0.823916\n",
      "Cost : 0.809592\n",
      "Cost : 0.801803\n",
      "Cost : 0.793557\n",
      "Cost : 0.787296\n",
      "Cost : 0.789329\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " ---- Train Accuracy ---- : 0.766542\n",
      " ---- Dev Accuracy ---- : 0.766667\n",
      "*** Fold Number ***1 ---\n",
      "*** Model Number ***6 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 11.0, 10.0, 10.0, 7.0, 3.0, 1]\n",
      "Cost : 0.730875\n",
      "Cost : 0.679977\n",
      "Cost : 0.675947\n",
      "Cost : 0.666465\n",
      "Cost : 0.68159\n",
      "Cost : 0.680871\n",
      "Cost : 0.666906\n",
      "Cost : 0.683099\n",
      "Cost : 0.676776\n",
      "Cost : 0.665995\n",
      " ---- Train Accuracy ---- : 0.612219\n",
      " ---- Dev Accuracy ---- : 0.651685\n",
      "*** Fold Number ***2 ---\n",
      "*** Model Number ***6 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 11.0, 10.0, 10.0, 7.0, 3.0, 1]\n",
      "Cost : 0.72492\n",
      "Cost : 0.672776\n",
      "Cost : 0.67278\n",
      "Cost : 0.665003\n",
      "Cost : 0.681347\n",
      "Cost : 0.679879\n",
      "Cost : 0.662383\n",
      "Cost : 0.681511\n",
      "Cost : 0.676663\n",
      "Cost : 0.663603\n",
      " ---- Train Accuracy ---- : 0.610973\n",
      " ---- Dev Accuracy ---- : 0.662921\n",
      "*** Fold Number ***3 ---\n",
      "*** Model Number ***6 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 11.0, 10.0, 10.0, 7.0, 3.0, 1]\n",
      "Cost : 0.716583\n",
      "Cost : 0.673529\n",
      "Cost : 0.668643\n",
      "Cost : 0.666354\n",
      "Cost : 0.677932\n",
      "Cost : 0.678413\n",
      "Cost : 0.663417\n",
      "Cost : 0.675261\n",
      "Cost : 0.674821\n",
      "Cost : 0.655805\n",
      " ---- Train Accuracy ---- : 0.617207\n",
      " ---- Dev Accuracy ---- : 0.606742\n",
      "*** Fold Number ***4 ---\n",
      "*** Model Number ***6 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 11.0, 10.0, 10.0, 7.0, 3.0, 1]\n",
      "Cost : 0.718751\n",
      "Cost : 0.677423\n",
      "Cost : 0.661611\n",
      "Cost : 0.664432\n",
      "Cost : 0.671119\n",
      "Cost : 0.694397\n",
      "Cost : 0.666346\n",
      "Cost : 0.677765\n",
      "Cost : 0.670993\n",
      "Cost : 0.664411\n",
      " ---- Train Accuracy ---- : 0.620948\n",
      " ---- Dev Accuracy ---- : 0.573034\n",
      "*** Fold Number ***5 ---\n",
      "*** Model Number ***6 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 11.0, 10.0, 10.0, 7.0, 3.0, 1]\n",
      "Cost : 0.708997\n",
      "Cost : 0.680101\n",
      "Cost : 0.665313\n",
      "Cost : 0.664119\n",
      "Cost : 0.673376\n",
      "Cost : 0.69942\n",
      "Cost : 0.66856\n",
      "Cost : 0.683165\n",
      "Cost : 0.672236\n",
      "Cost : 0.658533\n",
      " ---- Train Accuracy ---- : 0.617207\n",
      " ---- Dev Accuracy ---- : 0.606742\n",
      "*** Fold Number ***6 ---\n",
      "*** Model Number ***6 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 11.0, 10.0, 10.0, 7.0, 3.0, 1]\n",
      "Cost : 0.725848\n",
      "Cost : 0.682022\n",
      "Cost : 0.671052\n",
      "Cost : 0.667516\n",
      "Cost : 0.672826\n",
      "Cost : 0.69898\n",
      "Cost : 0.670446\n",
      "Cost : 0.69537\n",
      "Cost : 0.670205\n",
      "Cost : 0.667805\n",
      " ---- Train Accuracy ---- : 0.613466\n",
      " ---- Dev Accuracy ---- : 0.640449\n",
      "*** Fold Number ***7 ---\n",
      "*** Model Number ***6 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 11.0, 10.0, 10.0, 7.0, 3.0, 1]\n",
      "Cost : 0.732279\n",
      "Cost : 0.679302\n",
      "Cost : 0.662719\n",
      "Cost : 0.66259\n",
      "Cost : 0.667809\n",
      "Cost : 0.691383\n",
      "Cost : 0.663398\n",
      "Cost : 0.690281\n",
      "Cost : 0.669681\n",
      "Cost : 0.665914\n",
      " ---- Train Accuracy ---- : 0.619701\n",
      " ---- Dev Accuracy ---- : 0.58427\n",
      "*** Fold Number ***8 ---\n",
      "*** Model Number ***6 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 11.0, 10.0, 10.0, 7.0, 3.0, 1]\n",
      "Cost : 0.723862\n",
      "Cost : 0.682925\n",
      "Cost : 0.663528\n",
      "Cost : 0.660524\n",
      "Cost : 0.669986\n",
      "Cost : 0.70015\n",
      "Cost : 0.672751\n",
      "Cost : 0.691747\n",
      "Cost : 0.669428\n",
      "Cost : 0.669036\n",
      " ---- Train Accuracy ---- : 0.617207\n",
      " ---- Dev Accuracy ---- : 0.606742\n",
      "*** Fold Number ***9 ---\n",
      "*** Model Number ***6 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 11.0, 10.0, 10.0, 7.0, 3.0, 1]\n",
      "Cost : 0.72293\n",
      "Cost : 0.684311\n",
      "Cost : 0.665728\n",
      "Cost : 0.661476\n",
      "Cost : 0.669332\n",
      "Cost : 0.699494\n",
      "Cost : 0.671948\n",
      "Cost : 0.683296\n",
      "Cost : 0.674385\n",
      "Cost : 0.668081\n",
      " ---- Train Accuracy ---- : 0.61596\n",
      " ---- Dev Accuracy ---- : 0.617977\n",
      "*** Fold Number ***10 ---\n",
      "*** Model Number ***6 ---\n",
      "(1, 801)\n",
      "layers_dim [51, 11.0, 10.0, 10.0, 7.0, 3.0, 1]\n",
      "Cost : 0.756716\n",
      "Cost : 0.686699\n",
      "Cost : 0.681541\n",
      "Cost : 0.684809\n",
      "Cost : 0.670443\n",
      "Cost : 0.671011\n",
      "Cost : 0.676228\n",
      "Cost : 0.673811\n",
      "Cost : 0.677931\n",
      "Cost : 0.662123\n",
      " ---- Train Accuracy ---- : 0.616729\n",
      " ---- Dev Accuracy ---- : 0.611111\n",
      "*** Fold Number ***1 ---\n",
      "*** Model Number ***7 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 19.0, 12.0, 3.0, 1]\n",
      "Cost : 2.83849\n",
      "Cost : 2.48346\n",
      "Cost : 2.48414\n",
      "Cost : 2.47779\n",
      "Cost : 2.48505\n",
      "Cost : 2.47486\n",
      "Cost : 2.48035\n",
      "Cost : 2.48445\n",
      "Cost : 2.4761\n",
      "Cost : 2.48306\n",
      " ---- Train Accuracy ---- : 0.417706\n",
      " ---- Dev Accuracy ---- : 0.337079\n",
      "*** Fold Number ***2 ---\n",
      "*** Model Number ***7 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 19.0, 12.0, 3.0, 1]\n",
      "Cost : 2.8322\n",
      "Cost : 2.48133\n",
      "Cost : 2.47975\n",
      "Cost : 2.4761\n",
      "Cost : 2.48137\n",
      "Cost : 2.47538\n",
      "Cost : 2.47882\n",
      "Cost : 2.47947\n",
      "Cost : 2.47588\n",
      "Cost : 2.48427\n",
      " ---- Train Accuracy ---- : 0.416459\n",
      " ---- Dev Accuracy ---- : 0.370787\n",
      "*** Fold Number ***3 ---\n",
      "*** Model Number ***7 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 19.0, 12.0, 3.0, 1]\n",
      "Cost : 2.83518\n",
      "Cost : 2.48311\n",
      "Cost : 2.48143\n",
      "Cost : 2.47993\n",
      "Cost : 2.48647\n",
      "Cost : 2.48078\n",
      "Cost : 2.4842\n",
      "Cost : 2.48149\n",
      "Cost : 2.47591\n",
      "Cost : 2.48327\n",
      " ---- Train Accuracy ---- : 0.407731\n",
      " ---- Dev Accuracy ---- : 0.438202\n",
      "*** Fold Number ***4 ---\n",
      "*** Model Number ***7 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 19.0, 12.0, 3.0, 1]\n",
      "Cost : 2.83427\n",
      "Cost : 2.48161\n",
      "Cost : 2.47682\n",
      "Cost : 2.47939\n",
      "Cost : 2.4849\n",
      "Cost : 2.48148\n",
      "Cost : 2.48389\n",
      "Cost : 2.48128\n",
      "Cost : 2.47615\n",
      "Cost : 2.47911\n",
      " ---- Train Accuracy ---- : 0.408978\n",
      " ---- Dev Accuracy ---- : 0.460674\n",
      "*** Fold Number ***5 ---\n",
      "*** Model Number ***7 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 19.0, 12.0, 3.0, 1]\n",
      "Cost : 2.8327\n",
      "Cost : 2.48575\n",
      "Cost : 2.47452\n",
      "Cost : 2.47809\n",
      "Cost : 2.48444\n",
      "Cost : 2.48491\n",
      "Cost : 2.48514\n",
      "Cost : 2.47911\n",
      "Cost : 2.47986\n",
      "Cost : 2.48068\n",
      " ---- Train Accuracy ---- : 0.411471\n",
      " ---- Dev Accuracy ---- : 0.41573\n",
      "*** Fold Number ***6 ---\n",
      "*** Model Number ***7 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 19.0, 12.0, 3.0, 1]\n",
      "Cost : 2.83448\n",
      "Cost : 2.48594\n",
      "Cost : 2.47069\n",
      "Cost : 2.47614\n",
      "Cost : 2.48305\n",
      "Cost : 2.48355\n",
      "Cost : 2.48094\n",
      "Cost : 2.48089\n",
      "Cost : 2.48158\n",
      "Cost : 2.47983\n",
      " ---- Train Accuracy ---- : 0.406484\n",
      " ---- Dev Accuracy ---- : 0.449438\n",
      "*** Fold Number ***7 ---\n",
      "*** Model Number ***7 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 19.0, 12.0, 3.0, 1]\n",
      "Cost : 2.82877\n",
      "Cost : 2.48681\n",
      "Cost : 2.47285\n",
      "Cost : 2.47106\n",
      "Cost : 2.48196\n",
      "Cost : 2.48494\n",
      "Cost : 2.48382\n",
      "Cost : 2.47827\n",
      "Cost : 2.48145\n",
      "Cost : 2.47637\n",
      " ---- Train Accuracy ---- : 0.406484\n",
      " ---- Dev Accuracy ---- : 0.449438\n",
      "*** Fold Number ***8 ---\n",
      "*** Model Number ***7 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 19.0, 12.0, 3.0, 1]\n",
      "Cost : 2.82661\n",
      "Cost : 2.48538\n",
      "Cost : 2.47734\n",
      "Cost : 2.47099\n",
      "Cost : 2.48036\n",
      "Cost : 2.48065\n",
      "Cost : 2.48144\n",
      "Cost : 2.47764\n",
      "Cost : 2.48106\n",
      "Cost : 2.47732\n",
      " ---- Train Accuracy ---- : 0.412718\n",
      " ---- Dev Accuracy ---- : 0.426966\n",
      "*** Fold Number ***9 ---\n",
      "*** Model Number ***7 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 19.0, 12.0, 3.0, 1]\n",
      "Cost : 2.83187\n",
      "Cost : 2.48411\n",
      "Cost : 2.47576\n",
      "Cost : 2.47094\n",
      "Cost : 2.47986\n",
      "Cost : 2.48247\n",
      "Cost : 2.48036\n",
      "Cost : 2.47562\n",
      "Cost : 2.4813\n",
      "Cost : 2.4755\n",
      " ---- Train Accuracy ---- : 0.413965\n",
      " ---- Dev Accuracy ---- : 0.393258\n",
      "*** Fold Number ***10 ---\n",
      "*** Model Number ***7 ---\n",
      "(1, 801)\n",
      "layers_dim [51, 19.0, 12.0, 3.0, 1]\n",
      "Cost : 2.82605\n",
      "Cost : 2.48851\n",
      "Cost : 2.49016\n",
      "Cost : 2.48741\n",
      "Cost : 2.48837\n",
      "Cost : 2.48119\n",
      "Cost : 2.48465\n",
      "Cost : 2.47926\n",
      "Cost : 2.47694\n",
      "Cost : 2.48082\n",
      " ---- Train Accuracy ---- : 0.413233\n",
      " ---- Dev Accuracy ---- : 0.4\n",
      "*** Fold Number ***1 ---\n",
      "*** Model Number ***8 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 4.0, 4.0, 3.0, 2.0, 2.0, 1]\n",
      "Cost : 1.32378\n",
      "Cost : 0.667745\n",
      "Cost : 0.667745\n",
      "Cost : 0.667745\n",
      "Cost : 0.667745\n",
      "Cost : 0.667745\n",
      "Cost : 0.667745\n",
      "Cost : 0.667745\n",
      "Cost : 0.667745\n",
      "Cost : 0.667745\n",
      " ---- Train Accuracy ---- : 0.612219\n",
      " ---- Dev Accuracy ---- : 0.651685\n",
      "*** Fold Number ***2 ---\n",
      "*** Model Number ***8 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 4.0, 4.0, 3.0, 2.0, 2.0, 1]\n",
      "Cost : 1.32378\n",
      "Cost : 0.668312\n",
      "Cost : 0.668312\n",
      "Cost : 0.668312\n",
      "Cost : 0.668312\n",
      "Cost : 0.668312\n",
      "Cost : 0.668312\n",
      "Cost : 0.668312\n",
      "Cost : 0.668312\n",
      "Cost : 0.668311\n",
      " ---- Train Accuracy ---- : 0.610973\n",
      " ---- Dev Accuracy ---- : 0.662921\n",
      "*** Fold Number ***3 ---\n",
      "*** Model Number ***8 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 4.0, 4.0, 3.0, 2.0, 2.0, 1]\n",
      "Cost : 1.32378\n",
      "Cost : 0.665415\n",
      "Cost : 0.665415\n",
      "Cost : 0.665415\n",
      "Cost : 0.665415\n",
      "Cost : 0.665415\n",
      "Cost : 0.665415\n",
      "Cost : 0.665415\n",
      "Cost : 0.665415\n",
      "Cost : 0.665416\n",
      " ---- Train Accuracy ---- : 0.617207\n",
      " ---- Dev Accuracy ---- : 0.606742\n",
      "*** Fold Number ***4 ---\n",
      "*** Model Number ***8 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 4.0, 4.0, 3.0, 2.0, 2.0, 1]\n",
      "Cost : 1.32378\n",
      "Cost : 0.663598\n",
      "Cost : 0.663598\n",
      "Cost : 0.663598\n",
      "Cost : 0.663598\n",
      "Cost : 0.663598\n",
      "Cost : 0.663598\n",
      "Cost : 0.663598\n",
      "Cost : 0.663598\n",
      "Cost : 0.663599\n",
      " ---- Train Accuracy ---- : 0.620948\n",
      " ---- Dev Accuracy ---- : 0.573034\n",
      "*** Fold Number ***5 ---\n",
      "*** Model Number ***8 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 4.0, 4.0, 3.0, 2.0, 2.0, 1]\n",
      "Cost : 1.32378\n",
      "Cost : 0.665415\n",
      "Cost : 0.665415\n",
      "Cost : 0.665415\n",
      "Cost : 0.665415\n",
      "Cost : 0.665415\n",
      "Cost : 0.665415\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cost : 0.665415\n",
      "Cost : 0.665415\n",
      "Cost : 0.665414\n",
      " ---- Train Accuracy ---- : 0.617207\n",
      " ---- Dev Accuracy ---- : 0.606742\n",
      "*** Fold Number ***6 ---\n",
      "*** Model Number ***8 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 4.0, 4.0, 3.0, 2.0, 2.0, 1]\n",
      "Cost : 1.32378\n",
      "Cost : 0.667173\n",
      "Cost : 0.667173\n",
      "Cost : 0.667173\n",
      "Cost : 0.667173\n",
      "Cost : 0.667173\n",
      "Cost : 0.667173\n",
      "Cost : 0.667173\n",
      "Cost : 0.667173\n",
      "Cost : 0.667171\n",
      " ---- Train Accuracy ---- : 0.613466\n",
      " ---- Dev Accuracy ---- : 0.640449\n",
      "*** Fold Number ***7 ---\n",
      "*** Model Number ***8 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 4.0, 4.0, 3.0, 2.0, 2.0, 1]\n",
      "Cost : 1.32378\n",
      "Cost : 0.66421\n",
      "Cost : 0.66421\n",
      "Cost : 0.66421\n",
      "Cost : 0.66421\n",
      "Cost : 0.66421\n",
      "Cost : 0.66421\n",
      "Cost : 0.66421\n",
      "Cost : 0.66421\n",
      "Cost : 0.66421\n",
      " ---- Train Accuracy ---- : 0.619701\n",
      " ---- Dev Accuracy ---- : 0.58427\n",
      "*** Fold Number ***8 ---\n",
      "*** Model Number ***8 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 4.0, 4.0, 3.0, 2.0, 2.0, 1]\n",
      "Cost : 1.32378\n",
      "Cost : 0.665415\n",
      "Cost : 0.665415\n",
      "Cost : 0.665415\n",
      "Cost : 0.665415\n",
      "Cost : 0.665415\n",
      "Cost : 0.665415\n",
      "Cost : 0.665415\n",
      "Cost : 0.665414\n",
      "Cost : 0.665415\n",
      " ---- Train Accuracy ---- : 0.617207\n",
      " ---- Dev Accuracy ---- : 0.606742\n",
      "*** Fold Number ***9 ---\n",
      "*** Model Number ***8 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 4.0, 4.0, 3.0, 2.0, 2.0, 1]\n",
      "Cost : 1.32378\n",
      "Cost : 0.666007\n",
      "Cost : 0.666007\n",
      "Cost : 0.666007\n",
      "Cost : 0.666007\n",
      "Cost : 0.666007\n",
      "Cost : 0.666007\n",
      "Cost : 0.666007\n",
      "Cost : 0.666007\n",
      "Cost : 0.666007\n",
      " ---- Train Accuracy ---- : 0.61596\n",
      " ---- Dev Accuracy ---- : 0.617977\n",
      "*** Fold Number ***10 ---\n",
      "*** Model Number ***8 ---\n",
      "(1, 801)\n",
      "layers_dim [51, 20.0, 4.0, 4.0, 3.0, 2.0, 2.0, 1]\n",
      "Cost : 1.32378\n",
      "Cost : 0.665642\n",
      "Cost : 0.665642\n",
      "Cost : 0.665642\n",
      "Cost : 0.665642\n",
      "Cost : 0.665642\n",
      "Cost : 0.665642\n",
      "Cost : 0.665642\n",
      "Cost : 0.665642\n",
      "Cost : 0.665641\n",
      " ---- Train Accuracy ---- : 0.616729\n",
      " ---- Dev Accuracy ---- : 0.611111\n",
      "*** Fold Number ***1 ---\n",
      "*** Model Number ***9 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 11.0, 9.0, 8.0, 6.0, 5.0, 1]\n",
      "Cost : 1.08907\n",
      "Cost : 0.603617\n",
      "Cost : 0.561524\n",
      "Cost : 0.537604\n",
      "Cost : 0.519533\n",
      "Cost : 0.528026\n",
      "Cost : 0.531103\n",
      "Cost : 0.542697\n",
      "Cost : 0.543304\n",
      "Cost : 0.524983\n",
      " ---- Train Accuracy ---- : 0.846633\n",
      " ---- Dev Accuracy ---- : 0.820225\n",
      "*** Fold Number ***2 ---\n",
      "*** Model Number ***9 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 11.0, 9.0, 8.0, 6.0, 5.0, 1]\n",
      "Cost : 1.09106\n",
      "Cost : 0.597959\n",
      "Cost : 0.55708\n",
      "Cost : 0.528054\n",
      "Cost : 0.515965\n",
      "Cost : 0.519118\n",
      "Cost : 0.520154\n",
      "Cost : 0.530677\n",
      "Cost : 0.533283\n",
      "Cost : 0.517504\n",
      " ---- Train Accuracy ---- : 0.859102\n",
      " ---- Dev Accuracy ---- : 0.808989\n",
      "*** Fold Number ***3 ---\n",
      "*** Model Number ***9 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 11.0, 9.0, 8.0, 6.0, 5.0, 1]\n",
      "Cost : 1.09051\n",
      "Cost : 0.59507\n",
      "Cost : 0.556299\n",
      "Cost : 0.523866\n",
      "Cost : 0.507063\n",
      "Cost : 0.5255\n",
      "Cost : 0.514272\n",
      "Cost : 0.525802\n",
      "Cost : 0.537542\n",
      "Cost : 0.506691\n",
      " ---- Train Accuracy ---- : 0.857855\n",
      " ---- Dev Accuracy ---- : 0.764045\n",
      "*** Fold Number ***4 ---\n",
      "*** Model Number ***9 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 11.0, 9.0, 8.0, 6.0, 5.0, 1]\n",
      "Cost : 1.09021\n",
      "Cost : 0.589181\n",
      "Cost : 0.553047\n",
      "Cost : 0.522762\n",
      "Cost : 0.51834\n",
      "Cost : 0.52454\n",
      "Cost : 0.517485\n",
      "Cost : 0.524934\n",
      "Cost : 0.529945\n",
      "Cost : 0.514781\n",
      " ---- Train Accuracy ---- : 0.851621\n",
      " ---- Dev Accuracy ---- : 0.808989\n",
      "*** Fold Number ***5 ---\n",
      "*** Model Number ***9 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 11.0, 9.0, 8.0, 6.0, 5.0, 1]\n",
      "Cost : 1.09054\n",
      "Cost : 0.598788\n",
      "Cost : 0.555365\n",
      "Cost : 0.530212\n",
      "Cost : 0.524908\n",
      "Cost : 0.529794\n",
      "Cost : 0.521247\n",
      "Cost : 0.531417\n",
      "Cost : 0.532954\n",
      "Cost : 0.527021\n",
      " ---- Train Accuracy ---- : 0.851621\n",
      " ---- Dev Accuracy ---- : 0.842697\n",
      "*** Fold Number ***6 ---\n",
      "*** Model Number ***9 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 11.0, 9.0, 8.0, 6.0, 5.0, 1]\n",
      "Cost : 1.09007\n",
      "Cost : 0.602074\n",
      "Cost : 0.556143\n",
      "Cost : 0.532569\n",
      "Cost : 0.526187\n",
      "Cost : 0.535252\n",
      "Cost : 0.528995\n",
      "Cost : 0.542817\n",
      "Cost : 0.533094\n",
      "Cost : 0.532151\n",
      " ---- Train Accuracy ---- : 0.841646\n",
      " ---- Dev Accuracy ---- : 0.842697\n",
      "*** Fold Number ***7 ---\n",
      "*** Model Number ***9 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 11.0, 9.0, 8.0, 6.0, 5.0, 1]\n",
      "Cost : 1.09098\n",
      "Cost : 0.5923\n",
      "Cost : 0.544517\n",
      "Cost : 0.522372\n",
      "Cost : 0.525987\n",
      "Cost : 0.519276\n",
      "Cost : 0.510553\n",
      "Cost : 0.527268\n",
      "Cost : 0.518697\n",
      "Cost : 0.515748\n",
      " ---- Train Accuracy ---- : 0.861596\n",
      " ---- Dev Accuracy ---- : 0.831461\n",
      "*** Fold Number ***8 ---\n",
      "*** Model Number ***9 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 11.0, 9.0, 8.0, 6.0, 5.0, 1]\n",
      "Cost : 1.0912\n",
      "Cost : 0.583839\n",
      "Cost : 0.543939\n",
      "Cost : 0.522854\n",
      "Cost : 0.533423\n",
      "Cost : 0.50951\n",
      "Cost : 0.51773\n",
      "Cost : 0.52608\n",
      "Cost : 0.516277\n",
      "Cost : 0.522695\n",
      " ---- Train Accuracy ---- : 0.862843\n",
      " ---- Dev Accuracy ---- : 0.775281\n",
      "*** Fold Number ***9 ---\n",
      "*** Model Number ***9 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 11.0, 9.0, 8.0, 6.0, 5.0, 1]\n",
      "Cost : 1.09109\n",
      "Cost : 0.582936\n",
      "Cost : 0.54661\n",
      "Cost : 0.525102\n",
      "Cost : 0.537203\n",
      "Cost : 0.503084\n",
      "Cost : 0.514084\n",
      "Cost : 0.529972\n",
      "Cost : 0.505748\n",
      "Cost : 0.52223\n",
      " ---- Train Accuracy ---- : 0.861596\n",
      " ---- Dev Accuracy ---- : 0.786517\n",
      "*** Fold Number ***10 ---\n",
      "*** Model Number ***9 ---\n",
      "(1, 801)\n",
      "layers_dim [51, 11.0, 9.0, 8.0, 6.0, 5.0, 1]\n",
      "Cost : 1.09085\n",
      "Cost : 0.58681\n",
      "Cost : 0.545954\n",
      "Cost : 0.539297\n",
      "Cost : 0.510185\n",
      "Cost : 0.531033\n",
      "Cost : 0.526065\n",
      "Cost : 0.512966\n",
      "Cost : 0.530673\n",
      "Cost : 0.521563\n",
      " ---- Train Accuracy ---- : 0.855181\n",
      " ---- Dev Accuracy ---- : 0.833333\n",
      "*** Fold Number ***1 ---\n",
      "*** Model Number ***10 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 22.0, 9.0, 7.0, 3.0, 1]\n",
      "Cost : 0.890089\n",
      "Cost : 0.450333\n",
      "Cost : 0.434047\n",
      "Cost : 0.432226\n",
      "Cost : 0.429831\n",
      "Cost : 0.439846\n",
      "Cost : 0.437306\n",
      "Cost : 0.440847\n",
      "Cost : 0.41468\n",
      "Cost : 0.439739\n",
      " ---- Train Accuracy ---- : 0.875312\n",
      " ---- Dev Accuracy ---- : 0.865169\n",
      "*** Fold Number ***2 ---\n",
      "*** Model Number ***10 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 22.0, 9.0, 7.0, 3.0, 1]\n",
      "Cost : 0.890324\n",
      "Cost : 0.44448\n",
      "Cost : 0.415613\n",
      "Cost : 0.416168\n",
      "Cost : 0.41562\n",
      "Cost : 0.425959\n",
      "Cost : 0.428233\n",
      "Cost : 0.426222\n",
      "Cost : 0.406253\n",
      "Cost : 0.425121\n",
      " ---- Train Accuracy ---- : 0.88404\n",
      " ---- Dev Accuracy ---- : 0.797753\n",
      "*** Fold Number ***3 ---\n",
      "*** Model Number ***10 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 22.0, 9.0, 7.0, 3.0, 1]\n",
      "Cost : 0.890297\n",
      "Cost : 0.441075\n",
      "Cost : 0.428685\n",
      "Cost : 0.420772\n",
      "Cost : 0.420398\n",
      "Cost : 0.430937\n",
      "Cost : 0.439342\n",
      "Cost : 0.422476\n",
      "Cost : 0.411043\n",
      "Cost : 0.433168\n",
      " ---- Train Accuracy ---- : 0.872818\n",
      " ---- Dev Accuracy ---- : 0.775281\n",
      "*** Fold Number ***4 ---\n",
      "*** Model Number ***10 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 22.0, 9.0, 7.0, 3.0, 1]\n",
      "Cost : 0.890116\n",
      "Cost : 0.436844\n",
      "Cost : 0.421518\n",
      "Cost : 0.419557\n",
      "Cost : 0.416578\n",
      "Cost : 0.419361\n",
      "Cost : 0.432053\n",
      "Cost : 0.408925\n",
      "Cost : 0.412548\n",
      "Cost : 0.426409\n",
      " ---- Train Accuracy ---- : 0.885287\n",
      " ---- Dev Accuracy ---- : 0.786517\n",
      "*** Fold Number ***5 ---\n",
      "*** Model Number ***10 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 22.0, 9.0, 7.0, 3.0, 1]\n",
      "Cost : 0.890005\n",
      "Cost : 0.441009\n",
      "Cost : 0.419963\n",
      "Cost : 0.424699\n",
      "Cost : 0.412699\n",
      "Cost : 0.422732\n",
      "Cost : 0.433632\n",
      "Cost : 0.419888\n",
      "Cost : 0.414108\n",
      "Cost : 0.432766\n",
      " ---- Train Accuracy ---- : 0.882793\n",
      " ---- Dev Accuracy ---- : 0.808989\n",
      "*** Fold Number ***6 ---\n",
      "*** Model Number ***10 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 22.0, 9.0, 7.0, 3.0, 1]\n",
      "Cost : 0.88996\n",
      "Cost : 0.445181\n",
      "Cost : 0.432634\n",
      "Cost : 0.431798\n",
      "Cost : 0.43486\n",
      "Cost : 0.433939\n",
      "Cost : 0.441538\n",
      "Cost : 0.420782\n",
      "Cost : 0.422604\n",
      "Cost : 0.448227\n",
      " ---- Train Accuracy ---- : 0.876559\n",
      " ---- Dev Accuracy ---- : 0.876405\n",
      "*** Fold Number ***7 ---\n",
      "*** Model Number ***10 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 22.0, 9.0, 7.0, 3.0, 1]\n",
      "Cost : 0.890175\n",
      "Cost : 0.441975\n",
      "Cost : 0.415213\n",
      "Cost : 0.41254\n",
      "Cost : 0.414981\n",
      "Cost : 0.412574\n",
      "Cost : 0.430317\n",
      "Cost : 0.408495\n",
      "Cost : 0.4199\n",
      "Cost : 0.431427\n",
      " ---- Train Accuracy ---- : 0.891521\n",
      " ---- Dev Accuracy ---- : 0.831461\n",
      "*** Fold Number ***8 ---\n",
      "*** Model Number ***10 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 22.0, 9.0, 7.0, 3.0, 1]\n",
      "Cost : 0.889605\n",
      "Cost : 0.437291\n",
      "Cost : 0.415206\n",
      "Cost : 0.418335\n",
      "Cost : 0.422678\n",
      "Cost : 0.421533\n",
      "Cost : 0.435354\n",
      "Cost : 0.416036\n",
      "Cost : 0.428718\n",
      "Cost : 0.426837\n",
      " ---- Train Accuracy ---- : 0.882793\n",
      " ---- Dev Accuracy ---- : 0.786517\n",
      "*** Fold Number ***9 ---\n",
      "*** Model Number ***10 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 22.0, 9.0, 7.0, 3.0, 1]\n",
      "Cost : 0.890388\n",
      "Cost : 0.43954\n",
      "Cost : 0.415726\n",
      "Cost : 0.417965\n",
      "Cost : 0.421602\n",
      "Cost : 0.416484\n",
      "Cost : 0.42789\n",
      "Cost : 0.414312\n",
      "Cost : 0.43001\n",
      "Cost : 0.424906\n",
      " ---- Train Accuracy ---- : 0.882793\n",
      " ---- Dev Accuracy ---- : 0.808989\n",
      "*** Fold Number ***10 ---\n",
      "*** Model Number ***10 ---\n",
      "(1, 801)\n",
      "layers_dim [51, 22.0, 9.0, 7.0, 3.0, 1]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cost : 0.890576\n",
      "Cost : 0.418466\n",
      "Cost : 0.427799\n",
      "Cost : 0.418544\n",
      "Cost : 0.429086\n",
      "Cost : 0.427154\n",
      "Cost : 0.422992\n",
      "Cost : 0.430648\n",
      "Cost : 0.402857\n",
      "Cost : 0.433681\n",
      " ---- Train Accuracy ---- : 0.885144\n",
      " ---- Dev Accuracy ---- : 0.8\n",
      "*** Fold Number ***1 ---\n",
      "*** Model Number ***11 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 21.0, 14.0, 12.0, 11.0, 9.0, 3.0, 1]\n",
      "Cost : 0.824382\n",
      "Cost : 0.723928\n",
      "Cost : 0.679425\n",
      "Cost : 0.66261\n",
      "Cost : 0.644123\n",
      "Cost : 0.634663\n",
      "Cost : 0.662932\n",
      "Cost : 0.639952\n",
      "Cost : 0.634955\n",
      "Cost : 0.642847\n",
      " ---- Train Accuracy ---- : 0.820449\n",
      " ---- Dev Accuracy ---- : 0.853933\n",
      "*** Fold Number ***2 ---\n",
      "*** Model Number ***11 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 21.0, 14.0, 12.0, 11.0, 9.0, 3.0, 1]\n",
      "Cost : 0.825286\n",
      "Cost : 0.725825\n",
      "Cost : 0.679185\n",
      "Cost : 0.655105\n",
      "Cost : 0.637028\n",
      "Cost : 0.632177\n",
      "Cost : 0.647091\n",
      "Cost : 0.628787\n",
      "Cost : 0.625974\n",
      "Cost : 0.633874\n",
      " ---- Train Accuracy ---- : 0.82793\n",
      " ---- Dev Accuracy ---- : 0.842697\n",
      "*** Fold Number ***3 ---\n",
      "*** Model Number ***11 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 21.0, 14.0, 12.0, 11.0, 9.0, 3.0, 1]\n",
      "Cost : 0.830713\n",
      "Cost : 0.727759\n",
      "Cost : 0.682486\n",
      "Cost : 0.65368\n",
      "Cost : 0.638676\n",
      "Cost : 0.638767\n",
      "Cost : 0.648387\n",
      "Cost : 0.635835\n",
      "Cost : 0.623714\n",
      "Cost : 0.637232\n",
      " ---- Train Accuracy ---- : 0.82793\n",
      " ---- Dev Accuracy ---- : 0.820225\n",
      "*** Fold Number ***4 ---\n",
      "*** Model Number ***11 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 21.0, 14.0, 12.0, 11.0, 9.0, 3.0, 1]\n",
      "Cost : 0.833087\n",
      "Cost : 0.724529\n",
      "Cost : 0.683662\n",
      "Cost : 0.646799\n",
      "Cost : 0.633875\n",
      "Cost : 0.631474\n",
      "Cost : 0.647578\n",
      "Cost : 0.635615\n",
      "Cost : 0.615653\n",
      "Cost : 0.638904\n",
      " ---- Train Accuracy ---- : 0.829177\n",
      " ---- Dev Accuracy ---- : 0.786517\n",
      "*** Fold Number ***5 ---\n",
      "*** Model Number ***11 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 21.0, 14.0, 12.0, 11.0, 9.0, 3.0, 1]\n",
      "Cost : 0.835306\n",
      "Cost : 0.727085\n",
      "Cost : 0.689831\n",
      "Cost : 0.647006\n",
      "Cost : 0.636368\n",
      "Cost : 0.628715\n",
      "Cost : 0.651232\n",
      "Cost : 0.639922\n",
      "Cost : 0.625402\n",
      "Cost : 0.635751\n",
      " ---- Train Accuracy ---- : 0.830424\n",
      " ---- Dev Accuracy ---- : 0.797753\n",
      "*** Fold Number ***6 ---\n",
      "*** Model Number ***11 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 21.0, 14.0, 12.0, 11.0, 9.0, 3.0, 1]\n",
      "Cost : 0.835217\n",
      "Cost : 0.724972\n",
      "Cost : 0.692936\n",
      "Cost : 0.649776\n",
      "Cost : 0.642489\n",
      "Cost : 0.636197\n",
      "Cost : 0.647828\n",
      "Cost : 0.634857\n",
      "Cost : 0.629277\n",
      "Cost : 0.642417\n",
      " ---- Train Accuracy ---- : 0.821696\n",
      " ---- Dev Accuracy ---- : 0.853933\n",
      "*** Fold Number ***7 ---\n",
      "*** Model Number ***11 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 21.0, 14.0, 12.0, 11.0, 9.0, 3.0, 1]\n",
      "Cost : 0.835007\n",
      "Cost : 0.72244\n",
      "Cost : 0.690498\n",
      "Cost : 0.642866\n",
      "Cost : 0.639607\n",
      "Cost : 0.629883\n",
      "Cost : 0.642047\n",
      "Cost : 0.632781\n",
      "Cost : 0.62413\n",
      "Cost : 0.628193\n",
      " ---- Train Accuracy ---- : 0.829177\n",
      " ---- Dev Accuracy ---- : 0.775281\n",
      "*** Fold Number ***8 ---\n",
      "*** Model Number ***11 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 21.0, 14.0, 12.0, 11.0, 9.0, 3.0, 1]\n",
      "Cost : 0.83473\n",
      "Cost : 0.722088\n",
      "Cost : 0.682271\n",
      "Cost : 0.641026\n",
      "Cost : 0.640381\n",
      "Cost : 0.627679\n",
      "Cost : 0.635145\n",
      "Cost : 0.631188\n",
      "Cost : 0.62412\n",
      "Cost : 0.631205\n",
      " ---- Train Accuracy ---- : 0.826683\n",
      " ---- Dev Accuracy ---- : 0.797753\n",
      "*** Fold Number ***9 ---\n",
      "*** Model Number ***11 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 21.0, 14.0, 12.0, 11.0, 9.0, 3.0, 1]\n",
      "Cost : 0.835013\n",
      "Cost : 0.722687\n",
      "Cost : 0.685596\n",
      "Cost : 0.642789\n",
      "Cost : 0.639922\n",
      "Cost : 0.629451\n",
      "Cost : 0.642104\n",
      "Cost : 0.635418\n",
      "Cost : 0.62115\n",
      "Cost : 0.626733\n",
      " ---- Train Accuracy ---- : 0.830424\n",
      " ---- Dev Accuracy ---- : 0.797753\n",
      "*** Fold Number ***10 ---\n",
      "*** Model Number ***11 ---\n",
      "(1, 801)\n",
      "layers_dim [51, 21.0, 14.0, 12.0, 11.0, 9.0, 3.0, 1]\n",
      "Cost : 0.819113\n",
      "Cost : 0.729174\n",
      "Cost : 0.69551\n",
      "Cost : 0.657315\n",
      "Cost : 0.650784\n",
      "Cost : 0.622576\n",
      "Cost : 0.607881\n",
      "Cost : 0.630832\n",
      "Cost : 0.631696\n",
      "Cost : 0.628732\n",
      " ---- Train Accuracy ---- : 0.830212\n",
      " ---- Dev Accuracy ---- : 0.788889\n",
      "*** Fold Number ***1 ---\n",
      "*** Model Number ***12 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 18.0, 13.0, 11.0, 8.0, 1]\n",
      "Cost : 0.964661\n",
      "Cost : 0.877661\n",
      "Cost : 0.817158\n",
      "Cost : 0.792231\n",
      "Cost : 0.776873\n",
      "Cost : 0.759213\n",
      "Cost : 0.76508\n",
      "Cost : 0.749537\n",
      "Cost : 0.756972\n",
      "Cost : 0.765678\n",
      " ---- Train Accuracy ---- : 0.775561\n",
      " ---- Dev Accuracy ---- : 0.786517\n",
      "*** Fold Number ***2 ---\n",
      "*** Model Number ***12 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 18.0, 13.0, 11.0, 8.0, 1]\n",
      "Cost : 0.965467\n",
      "Cost : 0.874934\n",
      "Cost : 0.81237\n",
      "Cost : 0.789831\n",
      "Cost : 0.77302\n",
      "Cost : 0.7582\n",
      "Cost : 0.760475\n",
      "Cost : 0.746777\n",
      "Cost : 0.755386\n",
      "Cost : 0.762231\n",
      " ---- Train Accuracy ---- : 0.780549\n",
      " ---- Dev Accuracy ---- : 0.730337\n",
      "*** Fold Number ***3 ---\n",
      "*** Model Number ***12 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 18.0, 13.0, 11.0, 8.0, 1]\n",
      "Cost : 0.964698\n",
      "Cost : 0.875901\n",
      "Cost : 0.813928\n",
      "Cost : 0.786276\n",
      "Cost : 0.771059\n",
      "Cost : 0.757836\n",
      "Cost : 0.757813\n",
      "Cost : 0.743081\n",
      "Cost : 0.756484\n",
      "Cost : 0.762225\n",
      " ---- Train Accuracy ---- : 0.778055\n",
      " ---- Dev Accuracy ---- : 0.786517\n",
      "*** Fold Number ***4 ---\n",
      "*** Model Number ***12 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 18.0, 13.0, 11.0, 8.0, 1]\n",
      "Cost : 0.965441\n",
      "Cost : 0.874421\n",
      "Cost : 0.809212\n",
      "Cost : 0.785877\n",
      "Cost : 0.771139\n",
      "Cost : 0.753405\n",
      "Cost : 0.754871\n",
      "Cost : 0.734722\n",
      "Cost : 0.752419\n",
      "Cost : 0.761358\n",
      " ---- Train Accuracy ---- : 0.786783\n",
      " ---- Dev Accuracy ---- : 0.730337\n",
      "*** Fold Number ***5 ---\n",
      "*** Model Number ***12 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 18.0, 13.0, 11.0, 8.0, 1]\n",
      "Cost : 0.966615\n",
      "Cost : 0.873442\n",
      "Cost : 0.817569\n",
      "Cost : 0.789356\n",
      "Cost : 0.777435\n",
      "Cost : 0.757421\n",
      "Cost : 0.76092\n",
      "Cost : 0.748289\n",
      "Cost : 0.760523\n",
      "Cost : 0.764848\n",
      " ---- Train Accuracy ---- : 0.77182\n",
      " ---- Dev Accuracy ---- : 0.786517\n",
      "*** Fold Number ***6 ---\n",
      "*** Model Number ***12 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 18.0, 13.0, 11.0, 8.0, 1]\n",
      "Cost : 0.966658\n",
      "Cost : 0.875429\n",
      "Cost : 0.819404\n",
      "Cost : 0.790766\n",
      "Cost : 0.782226\n",
      "Cost : 0.761597\n",
      "Cost : 0.768638\n",
      "Cost : 0.75532\n",
      "Cost : 0.759835\n",
      "Cost : 0.767798\n",
      " ---- Train Accuracy ---- : 0.774314\n",
      " ---- Dev Accuracy ---- : 0.808989\n",
      "*** Fold Number ***7 ---\n",
      "*** Model Number ***12 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 18.0, 13.0, 11.0, 8.0, 1]\n",
      "Cost : 0.965978\n",
      "Cost : 0.86996\n",
      "Cost : 0.812637\n",
      "Cost : 0.788407\n",
      "Cost : 0.778961\n",
      "Cost : 0.755233\n",
      "Cost : 0.763106\n",
      "Cost : 0.748681\n",
      "Cost : 0.748401\n",
      "Cost : 0.758958\n",
      " ---- Train Accuracy ---- : 0.784289\n",
      " ---- Dev Accuracy ---- : 0.764045\n",
      "*** Fold Number ***8 ---\n",
      "*** Model Number ***12 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 18.0, 13.0, 11.0, 8.0, 1]\n",
      "Cost : 0.964339\n",
      "Cost : 0.870149\n",
      "Cost : 0.813892\n",
      "Cost : 0.789937\n",
      "Cost : 0.777479\n",
      "Cost : 0.758731\n",
      "Cost : 0.762497\n",
      "Cost : 0.748121\n",
      "Cost : 0.746069\n",
      "Cost : 0.762992\n",
      " ---- Train Accuracy ---- : 0.780549\n",
      " ---- Dev Accuracy ---- : 0.752809\n",
      "*** Fold Number ***9 ---\n",
      "*** Model Number ***12 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 20.0, 18.0, 13.0, 11.0, 8.0, 1]\n",
      "Cost : 0.96472\n",
      "Cost : 0.867212\n",
      "Cost : 0.812064\n",
      "Cost : 0.787697\n",
      "Cost : 0.778072\n",
      "Cost : 0.760527\n",
      "Cost : 0.759827\n",
      "Cost : 0.747387\n",
      "Cost : 0.749818\n",
      "Cost : 0.764354\n",
      " ---- Train Accuracy ---- : 0.784289\n",
      " ---- Dev Accuracy ---- : 0.752809\n",
      "*** Fold Number ***10 ---\n",
      "*** Model Number ***12 ---\n",
      "(1, 801)\n",
      "layers_dim [51, 20.0, 18.0, 13.0, 11.0, 8.0, 1]\n",
      "Cost : 0.963299\n",
      "Cost : 0.875067\n",
      "Cost : 0.819241\n",
      "Cost : 0.789094\n",
      "Cost : 0.760979\n",
      "Cost : 0.771923\n",
      "Cost : 0.765933\n",
      "Cost : 0.772029\n",
      "Cost : 0.753157\n",
      "Cost : 0.760261\n",
      " ---- Train Accuracy ---- : 0.779026\n",
      " ---- Dev Accuracy ---- : 0.788889\n",
      "*** Fold Number ***1 ---\n",
      "*** Model Number ***13 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 18.0, 18.0, 16.0, 12.0, 12.0, 1]\n",
      "Cost : 2.02809\n",
      "Cost : 1.61833\n",
      "Cost : 1.61773\n",
      "Cost : 1.61841\n",
      "Cost : 1.61818\n",
      "Cost : 1.61883\n",
      "Cost : 1.61919\n",
      "Cost : 1.6186\n",
      "Cost : 1.61685\n",
      "Cost : 1.6166\n",
      " ---- Train Accuracy ---- : 0.612219\n",
      " ---- Dev Accuracy ---- : 0.651685\n",
      "*** Fold Number ***2 ---\n",
      "*** Model Number ***13 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 18.0, 18.0, 16.0, 12.0, 12.0, 1]\n",
      "Cost : 2.02807\n",
      "Cost : 1.61868\n",
      "Cost : 1.61809\n",
      "Cost : 1.61838\n",
      "Cost : 1.6182\n",
      "Cost : 1.61854\n",
      "Cost : 1.61877\n",
      "Cost : 1.61911\n",
      "Cost : 1.61768\n",
      "Cost : 1.61698\n",
      " ---- Train Accuracy ---- : 0.610973\n",
      " ---- Dev Accuracy ---- : 0.662921\n",
      "*** Fold Number ***3 ---\n",
      "*** Model Number ***13 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 18.0, 18.0, 16.0, 12.0, 12.0, 1]\n",
      "Cost : 2.02795\n",
      "Cost : 1.61784\n",
      "Cost : 1.61722\n",
      "Cost : 1.61723\n",
      "Cost : 1.61737\n",
      "Cost : 1.61688\n",
      "Cost : 1.61733\n",
      "Cost : 1.61778\n",
      "Cost : 1.6173\n",
      "Cost : 1.61575\n",
      " ---- Train Accuracy ---- : 0.617207\n",
      " ---- Dev Accuracy ---- : 0.606742\n",
      "*** Fold Number ***4 ---\n",
      "*** Model Number ***13 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 18.0, 18.0, 16.0, 12.0, 12.0, 1]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cost : 2.02839\n",
      "Cost : 1.61628\n",
      "Cost : 1.61632\n",
      "Cost : 1.61662\n",
      "Cost : 1.61503\n",
      "Cost : 1.61479\n",
      "Cost : 1.61617\n",
      "Cost : 1.61628\n",
      "Cost : 1.61665\n",
      "Cost : 1.61402\n",
      " ---- Train Accuracy ---- : 0.620948\n",
      " ---- Dev Accuracy ---- : 0.573034\n",
      "*** Fold Number ***5 ---\n",
      "*** Model Number ***13 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 18.0, 18.0, 16.0, 12.0, 12.0, 1]\n",
      "Cost : 2.02735\n",
      "Cost : 1.61759\n",
      "Cost : 1.61707\n",
      "Cost : 1.61707\n",
      "Cost : 1.6154\n",
      "Cost : 1.61556\n",
      "Cost : 1.61688\n",
      "Cost : 1.61755\n",
      "Cost : 1.61844\n",
      "Cost : 1.61519\n",
      " ---- Train Accuracy ---- : 0.617207\n",
      " ---- Dev Accuracy ---- : 0.606742\n",
      "*** Fold Number ***6 ---\n",
      "*** Model Number ***13 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 18.0, 18.0, 16.0, 12.0, 12.0, 1]\n",
      "Cost : 2.02661\n",
      "Cost : 1.61825\n",
      "Cost : 1.61752\n",
      "Cost : 1.61731\n",
      "Cost : 1.61661\n",
      "Cost : 1.616\n",
      "Cost : 1.61761\n",
      "Cost : 1.61904\n",
      "Cost : 1.61865\n",
      "Cost : 1.61569\n",
      " ---- Train Accuracy ---- : 0.613466\n",
      " ---- Dev Accuracy ---- : 0.640449\n",
      "*** Fold Number ***7 ---\n",
      "*** Model Number ***13 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 18.0, 18.0, 16.0, 12.0, 12.0, 1]\n",
      "Cost : 2.02671\n",
      "Cost : 1.61696\n",
      "Cost : 1.61661\n",
      "Cost : 1.61598\n",
      "Cost : 1.61539\n",
      "Cost : 1.61477\n",
      "Cost : 1.61631\n",
      "Cost : 1.618\n",
      "Cost : 1.61731\n",
      "Cost : 1.61447\n",
      " ---- Train Accuracy ---- : 0.619701\n",
      " ---- Dev Accuracy ---- : 0.58427\n",
      "*** Fold Number ***8 ---\n",
      "*** Model Number ***13 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 18.0, 18.0, 16.0, 12.0, 12.0, 1]\n",
      "Cost : 2.02709\n",
      "Cost : 1.61734\n",
      "Cost : 1.61709\n",
      "Cost : 1.61615\n",
      "Cost : 1.61564\n",
      "Cost : 1.61567\n",
      "Cost : 1.61792\n",
      "Cost : 1.6192\n",
      "Cost : 1.61769\n",
      "Cost : 1.61612\n",
      " ---- Train Accuracy ---- : 0.617207\n",
      " ---- Dev Accuracy ---- : 0.606742\n",
      "*** Fold Number ***9 ---\n",
      "*** Model Number ***13 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 18.0, 18.0, 16.0, 12.0, 12.0, 1]\n",
      "Cost : 2.02708\n",
      "Cost : 1.61806\n",
      "Cost : 1.6174\n",
      "Cost : 1.61685\n",
      "Cost : 1.61577\n",
      "Cost : 1.61599\n",
      "Cost : 1.61754\n",
      "Cost : 1.61918\n",
      "Cost : 1.61898\n",
      "Cost : 1.61652\n",
      " ---- Train Accuracy ---- : 0.61596\n",
      " ---- Dev Accuracy ---- : 0.617977\n",
      "*** Fold Number ***10 ---\n",
      "*** Model Number ***13 ---\n",
      "(1, 801)\n",
      "layers_dim [51, 18.0, 18.0, 18.0, 16.0, 12.0, 12.0, 1]\n",
      "Cost : 2.02523\n",
      "Cost : 1.61803\n",
      "Cost : 1.61868\n",
      "Cost : 1.61655\n",
      "Cost : 1.61876\n",
      "Cost : 1.61753\n",
      "Cost : 1.61636\n",
      "Cost : 1.6167\n",
      "Cost : 1.616\n",
      "Cost : 1.61727\n",
      " ---- Train Accuracy ---- : 0.616729\n",
      " ---- Dev Accuracy ---- : 0.611111\n",
      "*** Fold Number ***1 ---\n",
      "*** Model Number ***14 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 22.0, 14.0, 12.0, 1]\n",
      "Cost : 1.15152\n",
      "Cost : 0.478423\n",
      "Cost : 0.470992\n",
      "Cost : 0.46749\n",
      "Cost : 0.468901\n",
      "Cost : 0.468465\n",
      "Cost : 0.477407\n",
      "Cost : 0.469693\n",
      "Cost : 0.465173\n",
      "Cost : 0.470804\n",
      " ---- Train Accuracy ---- : 0.855362\n",
      " ---- Dev Accuracy ---- : 0.853933\n",
      "*** Fold Number ***2 ---\n",
      "*** Model Number ***14 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 22.0, 14.0, 12.0, 1]\n",
      "Cost : 1.15269\n",
      "Cost : 0.467859\n",
      "Cost : 0.464508\n",
      "Cost : 0.454776\n",
      "Cost : 0.456834\n",
      "Cost : 0.454869\n",
      "Cost : 0.458329\n",
      "Cost : 0.461412\n",
      "Cost : 0.461263\n",
      "Cost : 0.458602\n",
      " ---- Train Accuracy ---- : 0.850374\n",
      " ---- Dev Accuracy ---- : 0.831461\n",
      "*** Fold Number ***3 ---\n",
      "*** Model Number ***14 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 22.0, 14.0, 12.0, 1]\n",
      "Cost : 1.15247\n",
      "Cost : 0.465391\n",
      "Cost : 0.459441\n",
      "Cost : 0.457099\n",
      "Cost : 0.459618\n",
      "Cost : 0.459393\n",
      "Cost : 0.448818\n",
      "Cost : 0.456811\n",
      "Cost : 0.456136\n",
      "Cost : 0.455739\n",
      " ---- Train Accuracy ---- : 0.86409\n",
      " ---- Dev Accuracy ---- : 0.797753\n",
      "*** Fold Number ***4 ---\n",
      "*** Model Number ***14 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 22.0, 14.0, 12.0, 1]\n",
      "Cost : 1.15528\n",
      "Cost : 0.468572\n",
      "Cost : 0.460508\n",
      "Cost : 0.457589\n",
      "Cost : 0.465524\n",
      "Cost : 0.452619\n",
      "Cost : 0.45037\n",
      "Cost : 0.458785\n",
      "Cost : 0.452734\n",
      "Cost : 0.460883\n",
      " ---- Train Accuracy ---- : 0.855362\n",
      " ---- Dev Accuracy ---- : 0.797753\n",
      "*** Fold Number ***5 ---\n",
      "*** Model Number ***14 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 22.0, 14.0, 12.0, 1]\n",
      "Cost : 1.15434\n",
      "Cost : 0.478682\n",
      "Cost : 0.472181\n",
      "Cost : 0.462895\n",
      "Cost : 0.474351\n",
      "Cost : 0.463589\n",
      "Cost : 0.4549\n",
      "Cost : 0.464841\n",
      "Cost : 0.460196\n",
      "Cost : 0.465283\n",
      " ---- Train Accuracy ---- : 0.856608\n",
      " ---- Dev Accuracy ---- : 0.865169\n",
      "*** Fold Number ***6 ---\n",
      "*** Model Number ***14 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 22.0, 14.0, 12.0, 1]\n",
      "Cost : 1.154\n",
      "Cost : 0.485171\n",
      "Cost : 0.47943\n",
      "Cost : 0.464194\n",
      "Cost : 0.480612\n",
      "Cost : 0.466357\n",
      "Cost : 0.458758\n",
      "Cost : 0.470401\n",
      "Cost : 0.466194\n",
      "Cost : 0.47199\n",
      " ---- Train Accuracy ---- : 0.851621\n",
      " ---- Dev Accuracy ---- : 0.865169\n",
      "*** Fold Number ***7 ---\n",
      "*** Model Number ***14 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 22.0, 14.0, 12.0, 1]\n",
      "Cost : 1.15335\n",
      "Cost : 0.473417\n",
      "Cost : 0.465904\n",
      "Cost : 0.451224\n",
      "Cost : 0.465758\n",
      "Cost : 0.453056\n",
      "Cost : 0.44575\n",
      "Cost : 0.457343\n",
      "Cost : 0.454272\n",
      "Cost : 0.458363\n",
      " ---- Train Accuracy ---- : 0.851621\n",
      " ---- Dev Accuracy ---- : 0.820225\n",
      "*** Fold Number ***8 ---\n",
      "*** Model Number ***14 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 22.0, 14.0, 12.0, 1]\n",
      "Cost : 1.15372\n",
      "Cost : 0.466697\n",
      "Cost : 0.465126\n",
      "Cost : 0.449678\n",
      "Cost : 0.465037\n",
      "Cost : 0.452708\n",
      "Cost : 0.447069\n",
      "Cost : 0.45386\n",
      "Cost : 0.449937\n",
      "Cost : 0.454835\n",
      " ---- Train Accuracy ---- : 0.856608\n",
      " ---- Dev Accuracy ---- : 0.797753\n",
      "*** Fold Number ***9 ---\n",
      "*** Model Number ***14 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 22.0, 14.0, 12.0, 1]\n",
      "Cost : 1.14924\n",
      "Cost : 0.471784\n",
      "Cost : 0.472202\n",
      "Cost : 0.455208\n",
      "Cost : 0.462028\n",
      "Cost : 0.45664\n",
      "Cost : 0.446761\n",
      "Cost : 0.458136\n",
      "Cost : 0.454117\n",
      "Cost : 0.457359\n",
      " ---- Train Accuracy ---- : 0.861596\n",
      " ---- Dev Accuracy ---- : 0.797753\n",
      "*** Fold Number ***10 ---\n",
      "*** Model Number ***14 ---\n",
      "(1, 801)\n",
      "layers_dim [51, 22.0, 14.0, 12.0, 1]\n",
      "Cost : 1.15108\n",
      "Cost : 0.468688\n",
      "Cost : 0.461235\n",
      "Cost : 0.449477\n",
      "Cost : 0.464807\n",
      "Cost : 0.466208\n",
      "Cost : 0.457239\n",
      "Cost : 0.458967\n",
      "Cost : 0.46894\n",
      "Cost : 0.467948\n",
      " ---- Train Accuracy ---- : 0.855181\n",
      " ---- Dev Accuracy ---- : 0.811111\n",
      "*** Fold Number ***1 ---\n",
      "*** Model Number ***15 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 17.0, 17.0, 12.0, 8.0, 2.0, 1]\n",
      "Cost : 0.838908\n",
      "Cost : 0.544129\n",
      "Cost : 0.537015\n",
      "Cost : 0.538791\n",
      "Cost : 0.538024\n",
      "Cost : 0.538746\n",
      "Cost : 0.537153\n",
      "Cost : 0.536213\n",
      "Cost : 0.540435\n",
      "Cost : 0.533952\n",
      " ---- Train Accuracy ---- : 0.859102\n",
      " ---- Dev Accuracy ---- : 0.865169\n",
      "*** Fold Number ***2 ---\n",
      "*** Model Number ***15 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 17.0, 17.0, 12.0, 8.0, 2.0, 1]\n",
      "Cost : 0.838361\n",
      "Cost : 0.534754\n",
      "Cost : 0.531488\n",
      "Cost : 0.529583\n",
      "Cost : 0.534702\n",
      "Cost : 0.533111\n",
      "Cost : 0.533517\n",
      "Cost : 0.531939\n",
      "Cost : 0.537123\n",
      "Cost : 0.529999\n",
      " ---- Train Accuracy ---- : 0.861596\n",
      " ---- Dev Accuracy ---- : 0.853933\n",
      "*** Fold Number ***3 ---\n",
      "*** Model Number ***15 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 17.0, 17.0, 12.0, 8.0, 2.0, 1]\n",
      "Cost : 0.839138\n",
      "Cost : 0.532588\n",
      "Cost : 0.529519\n",
      "Cost : 0.531822\n",
      "Cost : 0.537775\n",
      "Cost : 0.532594\n",
      "Cost : 0.534763\n",
      "Cost : 0.535417\n",
      "Cost : 0.534667\n",
      "Cost : 0.533177\n",
      " ---- Train Accuracy ---- : 0.86783\n",
      " ---- Dev Accuracy ---- : 0.808989\n",
      "*** Fold Number ***4 ---\n",
      "*** Model Number ***15 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 17.0, 17.0, 12.0, 8.0, 2.0, 1]\n",
      "Cost : 0.840143\n",
      "Cost : 0.537127\n",
      "Cost : 0.535507\n",
      "Cost : 0.535753\n",
      "Cost : 0.539715\n",
      "Cost : 0.537727\n",
      "Cost : 0.540125\n",
      "Cost : 0.539023\n",
      "Cost : 0.535966\n",
      "Cost : 0.53584\n",
      " ---- Train Accuracy ---- : 0.874065\n",
      " ---- Dev Accuracy ---- : 0.775281\n",
      "*** Fold Number ***5 ---\n",
      "*** Model Number ***15 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 17.0, 17.0, 12.0, 8.0, 2.0, 1]\n",
      "Cost : 0.840204\n",
      "Cost : 0.540493\n",
      "Cost : 0.542688\n",
      "Cost : 0.541056\n",
      "Cost : 0.545103\n",
      "Cost : 0.54181\n",
      "Cost : 0.542974\n",
      "Cost : 0.540128\n",
      "Cost : 0.543106\n",
      "Cost : 0.539086\n",
      " ---- Train Accuracy ---- : 0.862843\n",
      " ---- Dev Accuracy ---- : 0.808989\n",
      "*** Fold Number ***6 ---\n",
      "*** Model Number ***15 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 17.0, 17.0, 12.0, 8.0, 2.0, 1]\n",
      "Cost : 0.83975\n",
      "Cost : 0.542834\n",
      "Cost : 0.541409\n",
      "Cost : 0.543208\n",
      "Cost : 0.551015\n",
      "Cost : 0.54357\n",
      "Cost : 0.546067\n",
      "Cost : 0.543991\n",
      "Cost : 0.544999\n",
      "Cost : 0.544417\n",
      " ---- Train Accuracy ---- : 0.854115\n",
      " ---- Dev Accuracy ---- : 0.831461\n",
      "*** Fold Number ***7 ---\n",
      "*** Model Number ***15 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 17.0, 17.0, 12.0, 8.0, 2.0, 1]\n",
      "Cost : 0.840209\n",
      "Cost : 0.538753\n",
      "Cost : 0.539766\n",
      "Cost : 0.540249\n",
      "Cost : 0.546862\n",
      "Cost : 0.542286\n",
      "Cost : 0.542359\n",
      "Cost : 0.541189\n",
      "Cost : 0.544017\n",
      "Cost : 0.539854\n",
      " ---- Train Accuracy ---- : 0.861596\n",
      " ---- Dev Accuracy ---- : 0.808989\n",
      "*** Fold Number ***8 ---\n",
      "*** Model Number ***15 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 17.0, 17.0, 12.0, 8.0, 2.0, 1]\n",
      "Cost : 0.841254\n",
      "Cost : 0.536092\n",
      "Cost : 0.536392\n",
      "Cost : 0.536747\n",
      "Cost : 0.542498\n",
      "Cost : 0.537125\n",
      "Cost : 0.539796\n",
      "Cost : 0.537033\n",
      "Cost : 0.538105\n",
      "Cost : 0.536352\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " ---- Train Accuracy ---- : 0.86783\n",
      " ---- Dev Accuracy ---- : 0.797753\n",
      "*** Fold Number ***9 ---\n",
      "*** Model Number ***15 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 17.0, 17.0, 12.0, 8.0, 2.0, 1]\n",
      "Cost : 0.841558\n",
      "Cost : 0.535743\n",
      "Cost : 0.542061\n",
      "Cost : 0.540739\n",
      "Cost : 0.542548\n",
      "Cost : 0.540987\n",
      "Cost : 0.540753\n",
      "Cost : 0.537329\n",
      "Cost : 0.536898\n",
      "Cost : 0.540522\n",
      " ---- Train Accuracy ---- : 0.866584\n",
      " ---- Dev Accuracy ---- : 0.820225\n",
      "*** Fold Number ***10 ---\n",
      "*** Model Number ***15 ---\n",
      "(1, 801)\n",
      "layers_dim [51, 18.0, 17.0, 17.0, 12.0, 8.0, 2.0, 1]\n",
      "Cost : 0.840937\n",
      "Cost : 0.544101\n",
      "Cost : 0.540825\n",
      "Cost : 0.544788\n",
      "Cost : 0.534909\n",
      "Cost : 0.535307\n",
      "Cost : 0.541533\n",
      "Cost : 0.531791\n",
      "Cost : 0.540188\n",
      "Cost : 0.535376\n",
      " ---- Train Accuracy ---- : 0.867665\n",
      " ---- Dev Accuracy ---- : 0.822222\n",
      "*** Fold Number ***1 ---\n",
      "*** Model Number ***16 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 15.0, 5.0, 1]\n",
      "Cost : 1.14363\n",
      "Cost : 0.512245\n",
      "Cost : 0.505409\n",
      "Cost : 0.503457\n",
      "Cost : 0.502498\n",
      "Cost : 0.506108\n",
      "Cost : 0.503308\n",
      "Cost : 0.516985\n",
      "Cost : 0.502448\n",
      "Cost : 0.508788\n",
      " ---- Train Accuracy ---- : 0.830424\n",
      " ---- Dev Accuracy ---- : 0.820225\n",
      "*** Fold Number ***2 ---\n",
      "*** Model Number ***16 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 15.0, 5.0, 1]\n",
      "Cost : 1.14521\n",
      "Cost : 0.503713\n",
      "Cost : 0.496278\n",
      "Cost : 0.493605\n",
      "Cost : 0.496301\n",
      "Cost : 0.495494\n",
      "Cost : 0.492699\n",
      "Cost : 0.503548\n",
      "Cost : 0.494499\n",
      "Cost : 0.504353\n",
      " ---- Train Accuracy ---- : 0.830424\n",
      " ---- Dev Accuracy ---- : 0.808989\n",
      "*** Fold Number ***3 ---\n",
      "*** Model Number ***16 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 15.0, 5.0, 1]\n",
      "Cost : 1.1457\n",
      "Cost : 0.503422\n",
      "Cost : 0.496019\n",
      "Cost : 0.494514\n",
      "Cost : 0.497996\n",
      "Cost : 0.4934\n",
      "Cost : 0.497167\n",
      "Cost : 0.504136\n",
      "Cost : 0.49523\n",
      "Cost : 0.505533\n",
      " ---- Train Accuracy ---- : 0.829177\n",
      " ---- Dev Accuracy ---- : 0.797753\n",
      "*** Fold Number ***4 ---\n",
      "*** Model Number ***16 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 15.0, 5.0, 1]\n",
      "Cost : 1.14648\n",
      "Cost : 0.497766\n",
      "Cost : 0.49038\n",
      "Cost : 0.492865\n",
      "Cost : 0.493503\n",
      "Cost : 0.490629\n",
      "Cost : 0.492119\n",
      "Cost : 0.500586\n",
      "Cost : 0.492806\n",
      "Cost : 0.499454\n",
      " ---- Train Accuracy ---- : 0.836658\n",
      " ---- Dev Accuracy ---- : 0.786517\n",
      "*** Fold Number ***5 ---\n",
      "*** Model Number ***16 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 15.0, 5.0, 1]\n",
      "Cost : 1.14557\n",
      "Cost : 0.508812\n",
      "Cost : 0.501316\n",
      "Cost : 0.506191\n",
      "Cost : 0.500225\n",
      "Cost : 0.504793\n",
      "Cost : 0.502066\n",
      "Cost : 0.503738\n",
      "Cost : 0.50207\n",
      "Cost : 0.506618\n",
      " ---- Train Accuracy ---- : 0.830424\n",
      " ---- Dev Accuracy ---- : 0.865169\n",
      "*** Fold Number ***6 ---\n",
      "*** Model Number ***16 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 15.0, 5.0, 1]\n",
      "Cost : 1.14407\n",
      "Cost : 0.511695\n",
      "Cost : 0.506249\n",
      "Cost : 0.506135\n",
      "Cost : 0.50692\n",
      "Cost : 0.51024\n",
      "Cost : 0.506316\n",
      "Cost : 0.506313\n",
      "Cost : 0.504703\n",
      "Cost : 0.509792\n",
      " ---- Train Accuracy ---- : 0.821696\n",
      " ---- Dev Accuracy ---- : 0.820225\n",
      "*** Fold Number ***7 ---\n",
      "*** Model Number ***16 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 15.0, 5.0, 1]\n",
      "Cost : 1.14917\n",
      "Cost : 0.500713\n",
      "Cost : 0.49683\n",
      "Cost : 0.497193\n",
      "Cost : 0.498949\n",
      "Cost : 0.502124\n",
      "Cost : 0.494253\n",
      "Cost : 0.494536\n",
      "Cost : 0.498546\n",
      "Cost : 0.496698\n",
      " ---- Train Accuracy ---- : 0.829177\n",
      " ---- Dev Accuracy ---- : 0.808989\n",
      "*** Fold Number ***8 ---\n",
      "*** Model Number ***16 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 15.0, 5.0, 1]\n",
      "Cost : 1.14636\n",
      "Cost : 0.497477\n",
      "Cost : 0.497808\n",
      "Cost : 0.500288\n",
      "Cost : 0.498461\n",
      "Cost : 0.501941\n",
      "Cost : 0.494793\n",
      "Cost : 0.494935\n",
      "Cost : 0.499364\n",
      "Cost : 0.49015\n",
      " ---- Train Accuracy ---- : 0.82793\n",
      " ---- Dev Accuracy ---- : 0.808989\n",
      "*** Fold Number ***9 ---\n",
      "*** Model Number ***16 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 15.0, 5.0, 1]\n",
      "Cost : 1.14769\n",
      "Cost : 0.49569\n",
      "Cost : 0.50194\n",
      "Cost : 0.501362\n",
      "Cost : 0.501349\n",
      "Cost : 0.499331\n",
      "Cost : 0.496811\n",
      "Cost : 0.496465\n",
      "Cost : 0.50102\n",
      "Cost : 0.492009\n",
      " ---- Train Accuracy ---- : 0.835411\n",
      " ---- Dev Accuracy ---- : 0.842697\n",
      "*** Fold Number ***10 ---\n",
      "*** Model Number ***16 ---\n",
      "(1, 801)\n",
      "layers_dim [51, 15.0, 5.0, 1]\n",
      "Cost : 1.14367\n",
      "Cost : 0.512817\n",
      "Cost : 0.496586\n",
      "Cost : 0.497987\n",
      "Cost : 0.501309\n",
      "Cost : 0.49407\n",
      "Cost : 0.494242\n",
      "Cost : 0.505018\n",
      "Cost : 0.513215\n",
      "Cost : 0.500192\n",
      " ---- Train Accuracy ---- : 0.833958\n",
      " ---- Dev Accuracy ---- : 0.8\n",
      "*** Fold Number ***1 ---\n",
      "*** Model Number ***17 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 17.0, 16.0, 13.0, 8.0, 1]\n",
      "Cost : 4.45069\n",
      "Cost : 1.94689\n",
      "Cost : 1.67084\n",
      "Cost : 1.6197\n",
      "Cost : 1.60907\n",
      "Cost : 1.60775\n",
      "Cost : 1.60736\n",
      "Cost : 1.60785\n",
      "Cost : 1.60566\n",
      "Cost : 1.60818\n",
      " ---- Train Accuracy ---- : 0.612219\n",
      " ---- Dev Accuracy ---- : 0.651685\n",
      "*** Fold Number ***2 ---\n",
      "*** Model Number ***17 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 17.0, 16.0, 13.0, 8.0, 1]\n",
      "Cost : 4.45075\n",
      "Cost : 1.94746\n",
      "Cost : 1.67071\n",
      "Cost : 1.6196\n",
      "Cost : 1.60917\n",
      "Cost : 1.6083\n",
      "Cost : 1.60738\n",
      "Cost : 1.60812\n",
      "Cost : 1.60704\n",
      "Cost : 1.60873\n",
      " ---- Train Accuracy ---- : 0.610973\n",
      " ---- Dev Accuracy ---- : 0.662921\n",
      "*** Fold Number ***3 ---\n",
      "*** Model Number ***17 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 17.0, 16.0, 13.0, 8.0, 1]\n",
      "Cost : 4.4514\n",
      "Cost : 1.94546\n",
      "Cost : 1.66847\n",
      "Cost : 1.61756\n",
      "Cost : 1.6066\n",
      "Cost : 1.60577\n",
      "Cost : 1.60516\n",
      "Cost : 1.60424\n",
      "Cost : 1.6053\n",
      "Cost : 1.60561\n",
      " ---- Train Accuracy ---- : 0.617207\n",
      " ---- Dev Accuracy ---- : 0.606742\n",
      "*** Fold Number ***4 ---\n",
      "*** Model Number ***17 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 17.0, 16.0, 13.0, 8.0, 1]\n",
      "Cost : 4.45097\n",
      "Cost : 1.94404\n",
      "Cost : 1.66724\n",
      "Cost : 1.61644\n",
      "Cost : 1.6064\n",
      "Cost : 1.60515\n",
      "Cost : 1.60399\n",
      "Cost : 1.60353\n",
      "Cost : 1.6039\n",
      "Cost : 1.60372\n",
      " ---- Train Accuracy ---- : 0.620948\n",
      " ---- Dev Accuracy ---- : 0.573034\n",
      "*** Fold Number ***5 ---\n",
      "*** Model Number ***17 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 17.0, 16.0, 13.0, 8.0, 1]\n",
      "Cost : 4.45057\n",
      "Cost : 1.94547\n",
      "Cost : 1.66902\n",
      "Cost : 1.61697\n",
      "Cost : 1.60886\n",
      "Cost : 1.60608\n",
      "Cost : 1.60432\n",
      "Cost : 1.60497\n",
      "Cost : 1.60523\n",
      "Cost : 1.60537\n",
      " ---- Train Accuracy ---- : 0.617207\n",
      " ---- Dev Accuracy ---- : 0.606742\n",
      "*** Fold Number ***6 ---\n",
      "*** Model Number ***17 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 17.0, 16.0, 13.0, 8.0, 1]\n",
      "Cost : 4.45112\n",
      "Cost : 1.94733\n",
      "Cost : 1.67049\n",
      "Cost : 1.6182\n",
      "Cost : 1.61062\n",
      "Cost : 1.6063\n",
      "Cost : 1.60486\n",
      "Cost : 1.60632\n",
      "Cost : 1.60719\n",
      "Cost : 1.60727\n",
      " ---- Train Accuracy ---- : 0.613466\n",
      " ---- Dev Accuracy ---- : 0.640449\n",
      "*** Fold Number ***7 ---\n",
      "*** Model Number ***17 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 17.0, 16.0, 13.0, 8.0, 1]\n",
      "Cost : 4.45085\n",
      "Cost : 1.94473\n",
      "Cost : 1.66829\n",
      "Cost : 1.61734\n",
      "Cost : 1.60811\n",
      "Cost : 1.60396\n",
      "Cost : 1.60332\n",
      "Cost : 1.60432\n",
      "Cost : 1.60522\n",
      "Cost : 1.60634\n",
      " ---- Train Accuracy ---- : 0.619701\n",
      " ---- Dev Accuracy ---- : 0.58427\n",
      "*** Fold Number ***8 ---\n",
      "*** Model Number ***17 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 17.0, 16.0, 13.0, 8.0, 1]\n",
      "Cost : 4.45154\n",
      "Cost : 1.94657\n",
      "Cost : 1.6689\n",
      "Cost : 1.61789\n",
      "Cost : 1.60958\n",
      "Cost : 1.60541\n",
      "Cost : 1.60438\n",
      "Cost : 1.60567\n",
      "Cost : 1.60634\n",
      "Cost : 1.60595\n",
      " ---- Train Accuracy ---- : 0.617207\n",
      " ---- Dev Accuracy ---- : 0.606742\n",
      "*** Fold Number ***9 ---\n",
      "*** Model Number ***17 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 17.0, 16.0, 13.0, 8.0, 1]\n",
      "Cost : 4.45198\n",
      "Cost : 1.94661\n",
      "Cost : 1.66945\n",
      "Cost : 1.61746\n",
      "Cost : 1.60957\n",
      "Cost : 1.60541\n",
      "Cost : 1.60497\n",
      "Cost : 1.60586\n",
      "Cost : 1.607\n",
      "Cost : 1.6064\n",
      " ---- Train Accuracy ---- : 0.61596\n",
      " ---- Dev Accuracy ---- : 0.617977\n",
      "*** Fold Number ***10 ---\n",
      "*** Model Number ***17 ---\n",
      "(1, 801)\n",
      "layers_dim [51, 18.0, 17.0, 16.0, 13.0, 8.0, 1]\n",
      "Cost : 4.44811\n",
      "Cost : 1.94933\n",
      "Cost : 1.66956\n",
      "Cost : 1.61965\n",
      "Cost : 1.60722\n",
      "Cost : 1.60401\n",
      "Cost : 1.60743\n",
      "Cost : 1.60594\n",
      "Cost : 1.60715\n",
      "Cost : 1.60691\n",
      " ---- Train Accuracy ---- : 0.616729\n",
      " ---- Dev Accuracy ---- : 0.611111\n",
      "*** Fold Number ***1 ---\n",
      "*** Model Number ***18 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 1]\n",
      "Cost : 0.878801\n",
      "Cost : 0.791389\n",
      "Cost : 0.734917\n",
      "Cost : 0.692999\n",
      "Cost : 0.664822\n",
      "Cost : 0.642947\n",
      "Cost : 0.623957\n",
      "Cost : 0.615006\n",
      "Cost : 0.60262\n",
      "Cost : 0.591243\n",
      " ---- Train Accuracy ---- : 0.780549\n",
      " ---- Dev Accuracy ---- : 0.865169\n",
      "*** Fold Number ***2 ---\n",
      "*** Model Number ***18 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 1]\n",
      "Cost : 0.87736\n",
      "Cost : 0.788106\n",
      "Cost : 0.732966\n",
      "Cost : 0.688032\n",
      "Cost : 0.659256\n",
      "Cost : 0.636832\n",
      "Cost : 0.617557\n",
      "Cost : 0.608295\n",
      "Cost : 0.594308\n",
      "Cost : 0.585107\n",
      " ---- Train Accuracy ---- : 0.798005\n",
      " ---- Dev Accuracy ---- : 0.786517\n",
      "*** Fold Number ***3 ---\n",
      "*** Model Number ***18 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 1]\n",
      "Cost : 0.877612\n",
      "Cost : 0.787564\n",
      "Cost : 0.733637\n",
      "Cost : 0.68809\n",
      "Cost : 0.659708\n",
      "Cost : 0.63894\n",
      "Cost : 0.618221\n",
      "Cost : 0.609397\n",
      "Cost : 0.594726\n",
      "Cost : 0.585909\n",
      " ---- Train Accuracy ---- : 0.789277\n",
      " ---- Dev Accuracy ---- : 0.797753\n",
      "*** Fold Number ***4 ---\n",
      "*** Model Number ***18 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 1]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cost : 0.872263\n",
      "Cost : 0.779208\n",
      "Cost : 0.723995\n",
      "Cost : 0.679743\n",
      "Cost : 0.652246\n",
      "Cost : 0.629281\n",
      "Cost : 0.612083\n",
      "Cost : 0.600511\n",
      "Cost : 0.585249\n",
      "Cost : 0.576006\n",
      " ---- Train Accuracy ---- : 0.793017\n",
      " ---- Dev Accuracy ---- : 0.719101\n",
      "*** Fold Number ***5 ---\n",
      "*** Model Number ***18 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 1]\n",
      "Cost : 0.871529\n",
      "Cost : 0.780233\n",
      "Cost : 0.726462\n",
      "Cost : 0.682551\n",
      "Cost : 0.655392\n",
      "Cost : 0.634553\n",
      "Cost : 0.615686\n",
      "Cost : 0.603885\n",
      "Cost : 0.589909\n",
      "Cost : 0.580548\n",
      " ---- Train Accuracy ---- : 0.790524\n",
      " ---- Dev Accuracy ---- : 0.752809\n",
      "*** Fold Number ***6 ---\n",
      "*** Model Number ***18 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 1]\n",
      "Cost : 0.880381\n",
      "Cost : 0.789998\n",
      "Cost : 0.734541\n",
      "Cost : 0.690688\n",
      "Cost : 0.66614\n",
      "Cost : 0.642704\n",
      "Cost : 0.62556\n",
      "Cost : 0.611208\n",
      "Cost : 0.596639\n",
      "Cost : 0.587538\n",
      " ---- Train Accuracy ---- : 0.784289\n",
      " ---- Dev Accuracy ---- : 0.831461\n",
      "*** Fold Number ***7 ---\n",
      "*** Model Number ***18 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 1]\n",
      "Cost : 0.875867\n",
      "Cost : 0.78289\n",
      "Cost : 0.726142\n",
      "Cost : 0.682976\n",
      "Cost : 0.655651\n",
      "Cost : 0.633376\n",
      "Cost : 0.613889\n",
      "Cost : 0.601628\n",
      "Cost : 0.586345\n",
      "Cost : 0.577683\n",
      " ---- Train Accuracy ---- : 0.78803\n",
      " ---- Dev Accuracy ---- : 0.741573\n",
      "*** Fold Number ***8 ---\n",
      "*** Model Number ***18 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 1]\n",
      "Cost : 0.878877\n",
      "Cost : 0.785011\n",
      "Cost : 0.727077\n",
      "Cost : 0.686149\n",
      "Cost : 0.657757\n",
      "Cost : 0.634794\n",
      "Cost : 0.616098\n",
      "Cost : 0.604312\n",
      "Cost : 0.586552\n",
      "Cost : 0.579674\n",
      " ---- Train Accuracy ---- : 0.793017\n",
      " ---- Dev Accuracy ---- : 0.775281\n",
      "*** Fold Number ***9 ---\n",
      "*** Model Number ***18 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 18.0, 1]\n",
      "Cost : 0.874475\n",
      "Cost : 0.781515\n",
      "Cost : 0.724427\n",
      "Cost : 0.683812\n",
      "Cost : 0.656153\n",
      "Cost : 0.630588\n",
      "Cost : 0.614619\n",
      "Cost : 0.600251\n",
      "Cost : 0.583631\n",
      "Cost : 0.575568\n",
      " ---- Train Accuracy ---- : 0.791771\n",
      " ---- Dev Accuracy ---- : 0.764045\n",
      "*** Fold Number ***10 ---\n",
      "*** Model Number ***18 ---\n",
      "(1, 801)\n",
      "layers_dim [51, 18.0, 1]\n",
      "Cost : 0.88199\n",
      "Cost : 0.787501\n",
      "Cost : 0.728024\n",
      "Cost : 0.69048\n",
      "Cost : 0.663429\n",
      "Cost : 0.639833\n",
      "Cost : 0.618057\n",
      "Cost : 0.606514\n",
      "Cost : 0.597228\n",
      "Cost : 0.586566\n",
      " ---- Train Accuracy ---- : 0.787765\n",
      " ---- Dev Accuracy ---- : 0.777778\n",
      "*** Fold Number ***1 ---\n",
      "*** Model Number ***19 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 21.0, 18.0, 17.0, 10.0, 7.0, 4.0, 4.0, 1]\n",
      "Cost : 0.816751\n",
      "Cost : 0.377745\n",
      "Cost : 0.378074\n",
      "Cost : 0.376919\n",
      "Cost : 0.371458\n",
      "Cost : 0.384118\n",
      "Cost : 0.385931\n",
      "Cost : 0.382557\n",
      "Cost : 0.39181\n",
      "Cost : 0.379718\n",
      " ---- Train Accuracy ---- : 0.875312\n",
      " ---- Dev Accuracy ---- : 0.865169\n",
      "*** Fold Number ***2 ---\n",
      "*** Model Number ***19 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 21.0, 18.0, 17.0, 10.0, 7.0, 4.0, 4.0, 1]\n",
      "Cost : 0.817449\n",
      "Cost : 0.375756\n",
      "Cost : 0.380586\n",
      "Cost : 0.374195\n",
      "Cost : 0.376499\n",
      "Cost : 0.374229\n",
      "Cost : 0.387762\n",
      "Cost : 0.37521\n",
      "Cost : 0.384469\n",
      "Cost : 0.380877\n",
      " ---- Train Accuracy ---- : 0.876559\n",
      " ---- Dev Accuracy ---- : 0.842697\n",
      "*** Fold Number ***3 ---\n",
      "*** Model Number ***19 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 21.0, 18.0, 17.0, 10.0, 7.0, 4.0, 4.0, 1]\n",
      "Cost : 0.819482\n",
      "Cost : 0.361171\n",
      "Cost : 0.363981\n",
      "Cost : 0.357754\n",
      "Cost : 0.364965\n",
      "Cost : 0.352984\n",
      "Cost : 0.361053\n",
      "Cost : 0.362187\n",
      "Cost : 0.365804\n",
      "Cost : 0.364704\n",
      " ---- Train Accuracy ---- : 0.886534\n",
      " ---- Dev Accuracy ---- : 0.797753\n",
      "*** Fold Number ***4 ---\n",
      "*** Model Number ***19 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 21.0, 18.0, 17.0, 10.0, 7.0, 4.0, 4.0, 1]\n",
      "Cost : 0.816459\n",
      "Cost : 0.372616\n",
      "Cost : 0.36195\n",
      "Cost : 0.363498\n",
      "Cost : 0.369031\n",
      "Cost : 0.355868\n",
      "Cost : 0.371594\n",
      "Cost : 0.361039\n",
      "Cost : 0.371381\n",
      "Cost : 0.355822\n",
      " ---- Train Accuracy ---- : 0.88404\n",
      " ---- Dev Accuracy ---- : 0.764045\n",
      "*** Fold Number ***5 ---\n",
      "*** Model Number ***19 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 21.0, 18.0, 17.0, 10.0, 7.0, 4.0, 4.0, 1]\n",
      "Cost : 0.815544\n",
      "Cost : 0.36921\n",
      "Cost : 0.366687\n",
      "Cost : 0.363244\n",
      "Cost : 0.367317\n",
      "Cost : 0.36598\n",
      "Cost : 0.367078\n",
      "Cost : 0.365351\n",
      "Cost : 0.377653\n",
      "Cost : 0.36553\n",
      " ---- Train Accuracy ---- : 0.88404\n",
      " ---- Dev Accuracy ---- : 0.775281\n",
      "*** Fold Number ***6 ---\n",
      "*** Model Number ***19 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 21.0, 18.0, 17.0, 10.0, 7.0, 4.0, 4.0, 1]\n",
      "Cost : 0.817564\n",
      "Cost : 0.383842\n",
      "Cost : 0.38755\n",
      "Cost : 0.387234\n",
      "Cost : 0.399737\n",
      "Cost : 0.388666\n",
      "Cost : 0.384559\n",
      "Cost : 0.385839\n",
      "Cost : 0.392408\n",
      "Cost : 0.383373\n",
      " ---- Train Accuracy ---- : 0.871571\n",
      " ---- Dev Accuracy ---- : 0.876405\n",
      "*** Fold Number ***7 ---\n",
      "*** Model Number ***19 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 21.0, 18.0, 17.0, 10.0, 7.0, 4.0, 4.0, 1]\n",
      "Cost : 0.817808\n",
      "Cost : 0.36136\n",
      "Cost : 0.368364\n",
      "Cost : 0.363228\n",
      "Cost : 0.364212\n",
      "Cost : 0.359577\n",
      "Cost : 0.359483\n",
      "Cost : 0.358699\n",
      "Cost : 0.368786\n",
      "Cost : 0.354433\n",
      " ---- Train Accuracy ---- : 0.882793\n",
      " ---- Dev Accuracy ---- : 0.808989\n",
      "*** Fold Number ***8 ---\n",
      "*** Model Number ***19 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 21.0, 18.0, 17.0, 10.0, 7.0, 4.0, 4.0, 1]\n",
      "Cost : 0.817888\n",
      "Cost : 0.359616\n",
      "Cost : 0.359481\n",
      "Cost : 0.364388\n",
      "Cost : 0.373979\n",
      "Cost : 0.357398\n",
      "Cost : 0.363805\n",
      "Cost : 0.36635\n",
      "Cost : 0.368722\n",
      "Cost : 0.354699\n",
      " ---- Train Accuracy ---- : 0.879052\n",
      " ---- Dev Accuracy ---- : 0.808989\n",
      "*** Fold Number ***9 ---\n",
      "*** Model Number ***19 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 21.0, 18.0, 17.0, 10.0, 7.0, 4.0, 4.0, 1]\n",
      "Cost : 0.817652\n",
      "Cost : 0.364603\n",
      "Cost : 0.369561\n",
      "Cost : 0.368701\n",
      "Cost : 0.369579\n",
      "Cost : 0.363011\n",
      "Cost : 0.367678\n",
      "Cost : 0.36584\n",
      "Cost : 0.369538\n",
      "Cost : 0.364843\n",
      " ---- Train Accuracy ---- : 0.882793\n",
      " ---- Dev Accuracy ---- : 0.820225\n",
      "*** Fold Number ***10 ---\n",
      "*** Model Number ***19 ---\n",
      "(1, 801)\n",
      "layers_dim [51, 21.0, 18.0, 17.0, 10.0, 7.0, 4.0, 4.0, 1]\n",
      "Cost : 0.816024\n",
      "Cost : 0.375173\n",
      "Cost : 0.354977\n",
      "Cost : 0.362841\n",
      "Cost : 0.365033\n",
      "Cost : 0.369873\n",
      "Cost : 0.372235\n",
      "Cost : 0.358552\n",
      "Cost : 0.357478\n",
      "Cost : 0.361381\n",
      " ---- Train Accuracy ---- : 0.885144\n",
      " ---- Dev Accuracy ---- : 0.822222\n",
      "*** Fold Number ***1 ---\n",
      "*** Model Number ***20 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 17.0, 1]\n",
      "Cost : 1.04549\n",
      "Cost : 0.485585\n",
      "Cost : 0.488559\n",
      "Cost : 0.484859\n",
      "Cost : 0.487688\n",
      "Cost : 0.482414\n",
      "Cost : 0.48401\n",
      "Cost : 0.48481\n",
      "Cost : 0.48759\n",
      "Cost : 0.484881\n",
      " ---- Train Accuracy ---- : 0.817955\n",
      " ---- Dev Accuracy ---- : 0.820225\n",
      "*** Fold Number ***2 ---\n",
      "*** Model Number ***20 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 17.0, 1]\n",
      "Cost : 1.04406\n",
      "Cost : 0.476847\n",
      "Cost : 0.477187\n",
      "Cost : 0.475816\n",
      "Cost : 0.476409\n",
      "Cost : 0.473243\n",
      "Cost : 0.476013\n",
      "Cost : 0.47453\n",
      "Cost : 0.478031\n",
      "Cost : 0.478077\n",
      " ---- Train Accuracy ---- : 0.822943\n",
      " ---- Dev Accuracy ---- : 0.797753\n",
      "*** Fold Number ***3 ---\n",
      "*** Model Number ***20 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 17.0, 1]\n",
      "Cost : 1.04448\n",
      "Cost : 0.480342\n",
      "Cost : 0.480493\n",
      "Cost : 0.47703\n",
      "Cost : 0.479151\n",
      "Cost : 0.477695\n",
      "Cost : 0.478738\n",
      "Cost : 0.478177\n",
      "Cost : 0.480984\n",
      "Cost : 0.481329\n",
      " ---- Train Accuracy ---- : 0.820449\n",
      " ---- Dev Accuracy ---- : 0.808989\n",
      "*** Fold Number ***4 ---\n",
      "*** Model Number ***20 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 17.0, 1]\n",
      "Cost : 1.03825\n",
      "Cost : 0.474238\n",
      "Cost : 0.475634\n",
      "Cost : 0.469363\n",
      "Cost : 0.472428\n",
      "Cost : 0.471201\n",
      "Cost : 0.472548\n",
      "Cost : 0.4709\n",
      "Cost : 0.47746\n",
      "Cost : 0.474859\n",
      " ---- Train Accuracy ---- : 0.831671\n",
      " ---- Dev Accuracy ---- : 0.775281\n",
      "*** Fold Number ***5 ---\n",
      "*** Model Number ***20 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 17.0, 1]\n",
      "Cost : 1.0378\n",
      "Cost : 0.481749\n",
      "Cost : 0.484991\n",
      "Cost : 0.481489\n",
      "Cost : 0.481487\n",
      "Cost : 0.481698\n",
      "Cost : 0.480979\n",
      "Cost : 0.479784\n",
      "Cost : 0.485664\n",
      "Cost : 0.485033\n",
      " ---- Train Accuracy ---- : 0.819202\n",
      " ---- Dev Accuracy ---- : 0.853933\n",
      "*** Fold Number ***6 ---\n",
      "*** Model Number ***20 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 17.0, 1]\n",
      "Cost : 1.04683\n",
      "Cost : 0.48633\n",
      "Cost : 0.491283\n",
      "Cost : 0.487068\n",
      "Cost : 0.487134\n",
      "Cost : 0.486531\n",
      "Cost : 0.486069\n",
      "Cost : 0.486729\n",
      "Cost : 0.488526\n",
      "Cost : 0.487372\n",
      " ---- Train Accuracy ---- : 0.815461\n",
      " ---- Dev Accuracy ---- : 0.820225\n",
      "*** Fold Number ***7 ---\n",
      "*** Model Number ***20 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 17.0, 1]\n",
      "Cost : 1.0422\n",
      "Cost : 0.480885\n",
      "Cost : 0.479672\n",
      "Cost : 0.475538\n",
      "Cost : 0.476588\n",
      "Cost : 0.475381\n",
      "Cost : 0.475757\n",
      "Cost : 0.478453\n",
      "Cost : 0.479372\n",
      "Cost : 0.475878\n",
      " ---- Train Accuracy ---- : 0.821696\n",
      " ---- Dev Accuracy ---- : 0.797753\n",
      "*** Fold Number ***8 ---\n",
      "*** Model Number ***20 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 17.0, 1]\n",
      "Cost : 1.04446\n",
      "Cost : 0.480626\n",
      "Cost : 0.481129\n",
      "Cost : 0.475697\n",
      "Cost : 0.47658\n",
      "Cost : 0.476342\n",
      "Cost : 0.474751\n",
      "Cost : 0.479696\n",
      "Cost : 0.478168\n",
      "Cost : 0.475537\n",
      " ---- Train Accuracy ---- : 0.822943\n",
      " ---- Dev Accuracy ---- : 0.797753\n",
      "*** Fold Number ***9 ---\n",
      "*** Model Number ***20 ---\n",
      "(1, 802)\n",
      "layers_dim [51, 17.0, 1]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cost : 1.03985\n",
      "Cost : 0.484627\n",
      "Cost : 0.480905\n",
      "Cost : 0.477751\n",
      "Cost : 0.477837\n",
      "Cost : 0.475937\n",
      "Cost : 0.476243\n",
      "Cost : 0.480639\n",
      "Cost : 0.479458\n",
      "Cost : 0.47805\n",
      " ---- Train Accuracy ---- : 0.819202\n",
      " ---- Dev Accuracy ---- : 0.831461\n",
      "*** Fold Number ***10 ---\n",
      "*** Model Number ***20 ---\n",
      "(1, 801)\n",
      "layers_dim [51, 17.0, 1]\n",
      "Cost : 1.04659\n",
      "Cost : 0.486115\n",
      "Cost : 0.480804\n",
      "Cost : 0.486836\n",
      "Cost : 0.481171\n",
      "Cost : 0.482665\n",
      "Cost : 0.486393\n",
      "Cost : 0.481059\n",
      "Cost : 0.481365\n",
      "Cost : 0.479779\n",
      " ---- Train Accuracy ---- : 0.821473\n",
      " ---- Dev Accuracy ---- : 0.8\n",
      "        model_0_dev_acc  model_0_train_acc  model_1_dev_acc  \\\n",
      "fold                                                          \n",
      "fold_0         0.651685           0.612219         0.797753   \n",
      "fold_1         0.662921           0.610973         0.775281   \n",
      "fold_2         0.606742           0.617207         0.820225   \n",
      "fold_3         0.573034           0.620948         0.775281   \n",
      "fold_4         0.606742           0.617207         0.842697   \n",
      "fold_5         0.640449           0.613466         0.842697   \n",
      "fold_6         0.584270           0.619701         0.786517   \n",
      "fold_7         0.606742           0.617207         0.797753   \n",
      "fold_8         0.617977           0.615960         0.820225   \n",
      "fold_9         0.611111           0.616729         0.788889   \n",
      "mean           0.616167           0.616162         0.804732   \n",
      "std            0.025603           0.002841         0.022918   \n",
      "\n",
      "        model_1_train_acc  model_2_dev_acc  model_2_train_acc  \\\n",
      "fold                                                            \n",
      "fold_0           0.809227         0.876405           0.872818   \n",
      "fold_1           0.812968         0.853933           0.880299   \n",
      "fold_2           0.814214         0.808989           0.876559   \n",
      "fold_3           0.819202         0.764045           0.885287   \n",
      "fold_4           0.814214         0.820225           0.882793   \n",
      "fold_5           0.807980         0.853933           0.877805   \n",
      "fold_6           0.817955         0.808989           0.885287   \n",
      "fold_7           0.815461         0.808989           0.881546   \n",
      "fold_8           0.816708         0.820225           0.875312   \n",
      "fold_9           0.818976         0.811111           0.882647   \n",
      "mean             0.814691         0.822684           0.880035   \n",
      "std              0.003469         0.028582           0.003854   \n",
      "\n",
      "        model_3_dev_acc  model_3_train_acc  model_4_dev_acc  \\\n",
      "fold                                                          \n",
      "fold_0         0.651685           0.612219         0.842697   \n",
      "fold_1         0.662921           0.610973         0.797753   \n",
      "fold_2         0.606742           0.617207         0.752809   \n",
      "fold_3         0.573034           0.620948         0.651685   \n",
      "fold_4         0.606742           0.617207         0.775281   \n",
      "fold_5         0.640449           0.613466         0.831461   \n",
      "fold_6         0.584270           0.619701         0.741573   \n",
      "fold_7         0.606742           0.617207         0.752809   \n",
      "fold_8         0.617977           0.615960         0.752809   \n",
      "fold_9         0.611111           0.616729         0.766667   \n",
      "mean           0.616167           0.616162         0.766554   \n",
      "std            0.025603           0.002841         0.048096   \n",
      "\n",
      "        model_4_train_acc         ...          model_15_dev_acc  \\\n",
      "fold                              ...                             \n",
      "fold_0           0.763092         ...                  0.820225   \n",
      "fold_1           0.765586         ...                  0.808989   \n",
      "fold_2           0.768080         ...                  0.797753   \n",
      "fold_3           0.778055         ...                  0.786517   \n",
      "fold_4           0.763092         ...                  0.865169   \n",
      "fold_5           0.764339         ...                  0.820225   \n",
      "fold_6           0.768080         ...                  0.808989   \n",
      "fold_7           0.765586         ...                  0.808989   \n",
      "fold_8           0.763092         ...                  0.842697   \n",
      "fold_9           0.766542         ...                  0.800000   \n",
      "mean             0.766554         ...                  0.815955   \n",
      "std              0.004041         ...                  0.020827   \n",
      "\n",
      "        model_15_train_acc  model_16_dev_acc  model_16_train_acc  \\\n",
      "fold                                                               \n",
      "fold_0            0.830424          0.651685            0.612219   \n",
      "fold_1            0.830424          0.662921            0.610973   \n",
      "fold_2            0.829177          0.606742            0.617207   \n",
      "fold_3            0.836658          0.573034            0.620948   \n",
      "fold_4            0.830424          0.606742            0.617207   \n",
      "fold_5            0.821696          0.640449            0.613466   \n",
      "fold_6            0.829177          0.584270            0.619701   \n",
      "fold_7            0.827930          0.606742            0.617207   \n",
      "fold_8            0.835411          0.617977            0.615960   \n",
      "fold_9            0.833958          0.611111            0.616729   \n",
      "mean              0.830528          0.616167            0.616162   \n",
      "std               0.003833          0.025603            0.002841   \n",
      "\n",
      "        model_17_dev_acc  model_17_train_acc  model_18_dev_acc  \\\n",
      "fold                                                             \n",
      "fold_0          0.865169            0.780549          0.865169   \n",
      "fold_1          0.786517            0.798005          0.842697   \n",
      "fold_2          0.797753            0.789277          0.797753   \n",
      "fold_3          0.719101            0.793017          0.764045   \n",
      "fold_4          0.752809            0.790524          0.775281   \n",
      "fold_5          0.831461            0.784289          0.876405   \n",
      "fold_6          0.741573            0.788030          0.808989   \n",
      "fold_7          0.775281            0.793017          0.808989   \n",
      "fold_8          0.764045            0.791771          0.820225   \n",
      "fold_9          0.777778            0.787765          0.822222   \n",
      "mean            0.781149            0.789624          0.818177   \n",
      "std             0.038674            0.004427          0.032441   \n",
      "\n",
      "        model_18_train_acc  model_19_dev_acc  model_19_train_acc  \n",
      "fold                                                              \n",
      "fold_0            0.875312          0.820225            0.817955  \n",
      "fold_1            0.876559          0.797753            0.822943  \n",
      "fold_2            0.886534          0.808989            0.820449  \n",
      "fold_3            0.884040          0.775281            0.831671  \n",
      "fold_4            0.884040          0.853933            0.819202  \n",
      "fold_5            0.871571          0.820225            0.815461  \n",
      "fold_6            0.882793          0.797753            0.821696  \n",
      "fold_7            0.879052          0.797753            0.822943  \n",
      "fold_8            0.882793          0.831461            0.819202  \n",
      "fold_9            0.885144          0.800000            0.821473  \n",
      "mean              0.880784          0.810337            0.821299  \n",
      "std               0.004442          0.019897            0.003907  \n",
      "\n",
      "[12 rows x 40 columns]\n"
     ]
    }
   ],
   "source": [
    "## import best model\n",
    "best_hyper = pd.read_csv('output.csv')\n",
    "\n",
    "result_validate, random_hyper = cross_validate(x_train, y_train, num_fold = 10, num_random = 20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 117,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "decay               0.998091\n",
      "keep_prob           0.913567\n",
      "layers_dim       [13.0, 9.0]\n",
      "learning_rate    3.18624e-05\n",
      "scale_l2          0.00387048\n",
      "Name: model_14, dtype: object\n"
     ]
    }
   ],
   "source": [
    "print(random_hyper.loc['model_14',:])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "   A  B\n",
      "0  1  3\n",
      "1  2  4\n",
      "2  5  7\n",
      "3  6  8\n"
     ]
    }
   ],
   "source": [
    "k = pd.DataFrame({'A':[1,2],'B':[3,4]})\n",
    "t = pd.DataFrame({'A':[5,6],'B':[7,8]})\n",
    "\n",
    "gg = pd.concat([k, t], axis = 0, ignore_index = True)\n",
    "print(gg)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 182,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "result_validate.to_csv('result_validate.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 183,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "random_hyper.to_csv('result_hyper.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}