notebooks/03_classification.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "043933d1-0a38-49fe-b425-8f38e4f1f303",
   "metadata": {
    "tags": []
   },
   "source": [
    "# Exploring Temporal GNN Embeddings for Darknet Traffic Analysis\n",
    "## Final Classification Pipeline\n",
    "___\n",
    "\n",
    "## Table of Contents\n",
    "1. Main Classification Results\n",
    "2. Impact of History $H$\n",
    "3. Impact of Training Epochs\n",
    "4. Visualization\n",
    "\n",
    "This notebook contains the main codes to run the final k-Nearest-Neighbors classification pipeline. The main experiments are: (i) Main table with classification performance; (ii) impact of History parameter -- temporal aspect of tGNN; (iii) Impact of training epochs for incremental training."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "fc1950a4-a060-4872-8672-5d30b73711f6",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "from sklearn.metrics import classification_report\n",
    "\n",
    "import sys\n",
    "sys.path.append('../')\n",
    "\n",
    "from src.models.classification import KnnClassifier\n",
    "\n",
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4c909602-32ab-47d5-8cb1-c762bbaba1c9",
   "metadata": {},
   "source": [
    "The following code defines a function classification_pipeline that processes embeddings, performs classification using K-nearest neighbors (KNN) with Leave-One-Out validation, and returns a classification report."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "f6cefc16-ee8d-4053-ac46-d539a95f77ce",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load Ground Truth\n",
    "gt = pd.read_csv('../data/ground_truth/ground_truth.csv')\n",
    "\n",
    "# Used classes\n",
    "indices = ['mirai', 'shadowserver', 'driftnet', 'internetcensus', \n",
    "           'censys', 'rapid7', 'onyphe', 'netsystems', 'shodan', \n",
    "           'securitytrails', 'intrinsec', 'unk_bruteforcer', \n",
    "           'unk_spammer', 'unk_exploiter', 'macro avg']\n",
    "\n",
    "def classification_pipeline(embeddings, active, gt):\n",
    "    # Get active embeddings\n",
    "    embeddings = embeddings.loc[active]\n",
    "    # Manage column names and merge embeddings with labels\n",
    "    embeddings = embeddings.reset_index().rename(columns={'index':'src_ip'})\n",
    "    embeddings = embeddings.merge(gt, on='src_ip', how='left')\\\n",
    "                           .fillna('unknown')\\\n",
    "                           .set_index('src_ip')\n",
    "\n",
    "    # Take classes with more than 10 senders in the considered snapshot\n",
    "    tmp = embeddings.value_counts('label')\n",
    "    tmp = tmp[tmp>10].index\n",
    "    tmp_idx = embeddings[~embeddings.label.isin(tmp)].index\n",
    "    # Set the others as unknown\n",
    "    embeddings.loc[tmp_idx, 'label'] = 'unknown'\n",
    "    \n",
    "    # Manage X and y datasets\n",
    "    X_train = embeddings.drop(columns=['label']).values\n",
    "    y_train = np.ravel(embeddings[['label']].values)\n",
    "    \n",
    "    # Initialize the knn\n",
    "    knn = KnnClassifier(n_neighbors=3, metric='cosine')\n",
    "    knn.fit(X_train, y_train, scale_data=True)\n",
    "    \n",
    "    # Leave-One-Out validation\n",
    "    to_keep = np.where(y_train!='unknown')[0].reshape(-1, 1)\n",
    "    y_true = y_train[to_keep]\n",
    "    y_pred = knn.predict(to_keep, scale_data=True, loo=True) # Pass the indices\n",
    "\n",
    "    # Get the classification report\n",
    "    crep = classification_report(\n",
    "        y_true, \n",
    "        y_pred, \n",
    "        labels=np.unique(np.ravel(y_true)), \n",
    "        output_dict=True\n",
    "    )\n",
    "    \n",
    "    return crep"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7f6f08f5-de19-468d-b3b0-035bea6f744a",
   "metadata": {
    "tags": []
   },
   "source": [
    "## 1. Main Classification Results"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e93a12cc-f056-4fa0-8f9f-8aa4d05ab30a",
   "metadata": {},
   "source": [
    "The following code iterates through each trained model and performs various operations for each column:\n",
    "1. **Manage file paths** and names based on the column name.\n",
    "2. **Iterate through days** from 21 to 31: Load a list of active senders from a CSV file and retrieve embeddings from a CSV file. Run a classification pipeline and extracts F1-Scores, then store the F1-Scores in the `reports` list.\n",
    "3. **Calculate the mean and standard deviation** of F1-Scores.\n",
    "4. **Organize the results as a dataframe and appends it to the `final_reps` list.\n",
    "5. **Concatenate** all the dataframes in the `final_reps` list and saves the results to a CSV file.\n",
    "6. **Display** the resulting dataframe**."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "695dae7f-fe59-42ac-8e2b-df3cc36257c3",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "69522c61f91445e0b554c0b8685ea3d3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/11 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "0c87bfed2282450a9038b220622e21b6",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/11 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "817a7f590fe6405aaca679b17c628391",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/11 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "bbaf736d2dce4064bf481d17eb168374",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/11 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "0414cfd8ed604422b668a56e07618a05",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/11 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "816f053d74fd4feeaab064b2ed4c4f1d",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/11 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ed89fe8be09144efbd3adfe608e95875",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/11 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "78f287d0f7a94c98a03b87179b53dfd0",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/11 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "efb16d4a6d82458c9265a4b583e887b3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/11 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>iDarkVec</th>\n",
       "      <th>GCN</th>\n",
       "      <th>iGCN</th>\n",
       "      <th>GCNGRU</th>\n",
       "      <th>iGCNGRU</th>\n",
       "      <th>GCN_f</th>\n",
       "      <th>iGCN_f</th>\n",
       "      <th>GCNGRU_f</th>\n",
       "      <th>iGCNGRU_f</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>mirai</th>\n",
       "      <td>0.98$\\pm$0.00</td>\n",
       "      <td>0.63$\\pm$0.03</td>\n",
       "      <td>0.76$\\pm$0.02</td>\n",
       "      <td>0.69$\\pm$0.03</td>\n",
       "      <td>0.72$\\pm$0.02</td>\n",
       "      <td>0.98$\\pm$0.00</td>\n",
       "      <td>0.98$\\pm$0.00</td>\n",
       "      <td>0.88$\\pm$0.06</td>\n",
       "      <td>0.98$\\pm$0.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>shadowserver</th>\n",
       "      <td>0.96$\\pm$0.02</td>\n",
       "      <td>0.14$\\pm$0.13</td>\n",
       "      <td>0.49$\\pm$0.15</td>\n",
       "      <td>0.64$\\pm$0.07</td>\n",
       "      <td>0.49$\\pm$0.16</td>\n",
       "      <td>0.91$\\pm$0.01</td>\n",
       "      <td>0.90$\\pm$0.02</td>\n",
       "      <td>0.95$\\pm$0.01</td>\n",
       "      <td>0.96$\\pm$0.01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>driftnet</th>\n",
       "      <td>0.96$\\pm$0.01</td>\n",
       "      <td>0.77$\\pm$0.09</td>\n",
       "      <td>0.73$\\pm$0.06</td>\n",
       "      <td>0.87$\\pm$0.05</td>\n",
       "      <td>0.90$\\pm$0.02</td>\n",
       "      <td>0.88$\\pm$0.04</td>\n",
       "      <td>0.85$\\pm$0.05</td>\n",
       "      <td>0.97$\\pm$0.02</td>\n",
       "      <td>0.97$\\pm$0.02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>internetcensus</th>\n",
       "      <td>0.99$\\pm$0.00</td>\n",
       "      <td>0.26$\\pm$0.12</td>\n",
       "      <td>0.33$\\pm$0.08</td>\n",
       "      <td>0.68$\\pm$0.05</td>\n",
       "      <td>0.47$\\pm$0.13</td>\n",
       "      <td>0.64$\\pm$0.12</td>\n",
       "      <td>0.58$\\pm$0.14</td>\n",
       "      <td>0.90$\\pm$0.04</td>\n",
       "      <td>0.92$\\pm$0.02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>censys</th>\n",
       "      <td>0.94$\\pm$0.01</td>\n",
       "      <td>0.53$\\pm$0.06</td>\n",
       "      <td>0.60$\\pm$0.05</td>\n",
       "      <td>0.59$\\pm$0.05</td>\n",
       "      <td>0.60$\\pm$0.03</td>\n",
       "      <td>0.89$\\pm$0.03</td>\n",
       "      <td>0.88$\\pm$0.03</td>\n",
       "      <td>0.91$\\pm$0.02</td>\n",
       "      <td>0.92$\\pm$0.01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rapid7</th>\n",
       "      <td>1.00$\\pm$0.00</td>\n",
       "      <td>0.91$\\pm$0.05</td>\n",
       "      <td>0.76$\\pm$0.11</td>\n",
       "      <td>0.91$\\pm$0.02</td>\n",
       "      <td>0.83$\\pm$0.10</td>\n",
       "      <td>0.84$\\pm$0.04</td>\n",
       "      <td>0.84$\\pm$0.02</td>\n",
       "      <td>0.69$\\pm$0.47</td>\n",
       "      <td>0.97$\\pm$0.04</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>onyphe</th>\n",
       "      <td>0.96$\\pm$0.02</td>\n",
       "      <td>0.06$\\pm$0.05</td>\n",
       "      <td>0.05$\\pm$0.04</td>\n",
       "      <td>0.11$\\pm$0.06</td>\n",
       "      <td>0.05$\\pm$0.05</td>\n",
       "      <td>0.93$\\pm$0.03</td>\n",
       "      <td>0.93$\\pm$0.02</td>\n",
       "      <td>0.92$\\pm$0.01</td>\n",
       "      <td>0.91$\\pm$0.05</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>netsystems</th>\n",
       "      <td>0.45$\\pm$0.07</td>\n",
       "      <td>0.02$\\pm$0.03</td>\n",
       "      <td>0.09$\\pm$0.08</td>\n",
       "      <td>0.32$\\pm$0.05</td>\n",
       "      <td>0.06$\\pm$0.05</td>\n",
       "      <td>0.97$\\pm$0.03</td>\n",
       "      <td>0.97$\\pm$0.02</td>\n",
       "      <td>0.80$\\pm$0.09</td>\n",
       "      <td>0.95$\\pm$0.04</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>shodan</th>\n",
       "      <td>0.88$\\pm$0.02</td>\n",
       "      <td>0.12$\\pm$0.16</td>\n",
       "      <td>0.14$\\pm$0.16</td>\n",
       "      <td>0.23$\\pm$0.10</td>\n",
       "      <td>0.10$\\pm$0.14</td>\n",
       "      <td>0.81$\\pm$0.05</td>\n",
       "      <td>0.79$\\pm$0.06</td>\n",
       "      <td>0.76$\\pm$0.05</td>\n",
       "      <td>0.79$\\pm$0.06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>securitytrails</th>\n",
       "      <td>1.00$\\pm$0.00</td>\n",
       "      <td>0.12$\\pm$0.25</td>\n",
       "      <td>0.86$\\pm$0.07</td>\n",
       "      <td>0.87$\\pm$0.04</td>\n",
       "      <td>0.67$\\pm$0.04</td>\n",
       "      <td>0.98$\\pm$0.04</td>\n",
       "      <td>0.96$\\pm$0.05</td>\n",
       "      <td>0.97$\\pm$0.04</td>\n",
       "      <td>0.98$\\pm$0.03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>intrinsec</th>\n",
       "      <td>0.94$\\pm$0.02</td>\n",
       "      <td>0.00$\\pm$0.00</td>\n",
       "      <td>0.00$\\pm$0.00</td>\n",
       "      <td>0.00$\\pm$0.00</td>\n",
       "      <td>0.00$\\pm$0.00</td>\n",
       "      <td>0.32$\\pm$0.24</td>\n",
       "      <td>0.26$\\pm$0.19</td>\n",
       "      <td>0.75$\\pm$0.12</td>\n",
       "      <td>0.73$\\pm$0.13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>unk_bruteforcer</th>\n",
       "      <td>0.60$\\pm$0.04</td>\n",
       "      <td>0.10$\\pm$0.04</td>\n",
       "      <td>0.53$\\pm$0.03</td>\n",
       "      <td>0.53$\\pm$0.03</td>\n",
       "      <td>0.52$\\pm$0.01</td>\n",
       "      <td>0.62$\\pm$0.03</td>\n",
       "      <td>0.61$\\pm$0.03</td>\n",
       "      <td>0.61$\\pm$0.03</td>\n",
       "      <td>0.63$\\pm$0.03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>unk_spammer</th>\n",
       "      <td>0.42$\\pm$0.04</td>\n",
       "      <td>0.14$\\pm$0.04</td>\n",
       "      <td>0.23$\\pm$0.04</td>\n",
       "      <td>0.29$\\pm$0.04</td>\n",
       "      <td>0.19$\\pm$0.07</td>\n",
       "      <td>0.47$\\pm$0.06</td>\n",
       "      <td>0.46$\\pm$0.06</td>\n",
       "      <td>0.42$\\pm$0.08</td>\n",
       "      <td>0.47$\\pm$0.08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>unk_exploiter</th>\n",
       "      <td>0.11$\\pm$0.14</td>\n",
       "      <td>0.07$\\pm$0.20</td>\n",
       "      <td>0.07$\\pm$0.20</td>\n",
       "      <td>0.00$\\pm$0.00</td>\n",
       "      <td>0.06$\\pm$0.16</td>\n",
       "      <td>0.07$\\pm$0.20</td>\n",
       "      <td>0.07$\\pm$0.20</td>\n",
       "      <td>0.00$\\pm$0.00</td>\n",
       "      <td>0.10$\\pm$0.21</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>macro avg</th>\n",
       "      <td>0.78$\\pm$0.03</td>\n",
       "      <td>0.26$\\pm$0.05</td>\n",
       "      <td>0.41$\\pm$0.04</td>\n",
       "      <td>0.49$\\pm$0.04</td>\n",
       "      <td>0.41$\\pm$0.04</td>\n",
       "      <td>0.74$\\pm$0.02</td>\n",
       "      <td>0.73$\\pm$0.03</td>\n",
       "      <td>0.75$\\pm$0.03</td>\n",
       "      <td>0.80$\\pm$0.02</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                      iDarkVec            GCN           iGCN         GCNGRU  \\\n",
       "0                                                                             \n",
       "mirai            0.98$\\pm$0.00  0.63$\\pm$0.03  0.76$\\pm$0.02  0.69$\\pm$0.03   \n",
       "shadowserver     0.96$\\pm$0.02  0.14$\\pm$0.13  0.49$\\pm$0.15  0.64$\\pm$0.07   \n",
       "driftnet         0.96$\\pm$0.01  0.77$\\pm$0.09  0.73$\\pm$0.06  0.87$\\pm$0.05   \n",
       "internetcensus   0.99$\\pm$0.00  0.26$\\pm$0.12  0.33$\\pm$0.08  0.68$\\pm$0.05   \n",
       "censys           0.94$\\pm$0.01  0.53$\\pm$0.06  0.60$\\pm$0.05  0.59$\\pm$0.05   \n",
       "rapid7           1.00$\\pm$0.00  0.91$\\pm$0.05  0.76$\\pm$0.11  0.91$\\pm$0.02   \n",
       "onyphe           0.96$\\pm$0.02  0.06$\\pm$0.05  0.05$\\pm$0.04  0.11$\\pm$0.06   \n",
       "netsystems       0.45$\\pm$0.07  0.02$\\pm$0.03  0.09$\\pm$0.08  0.32$\\pm$0.05   \n",
       "shodan           0.88$\\pm$0.02  0.12$\\pm$0.16  0.14$\\pm$0.16  0.23$\\pm$0.10   \n",
       "securitytrails   1.00$\\pm$0.00  0.12$\\pm$0.25  0.86$\\pm$0.07  0.87$\\pm$0.04   \n",
       "intrinsec        0.94$\\pm$0.02  0.00$\\pm$0.00  0.00$\\pm$0.00  0.00$\\pm$0.00   \n",
       "unk_bruteforcer  0.60$\\pm$0.04  0.10$\\pm$0.04  0.53$\\pm$0.03  0.53$\\pm$0.03   \n",
       "unk_spammer      0.42$\\pm$0.04  0.14$\\pm$0.04  0.23$\\pm$0.04  0.29$\\pm$0.04   \n",
       "unk_exploiter    0.11$\\pm$0.14  0.07$\\pm$0.20  0.07$\\pm$0.20  0.00$\\pm$0.00   \n",
       "macro avg        0.78$\\pm$0.03  0.26$\\pm$0.05  0.41$\\pm$0.04  0.49$\\pm$0.04   \n",
       "\n",
       "                       iGCNGRU          GCN_f         iGCN_f       GCNGRU_f  \\\n",
       "0                                                                             \n",
       "mirai            0.72$\\pm$0.02  0.98$\\pm$0.00  0.98$\\pm$0.00  0.88$\\pm$0.06   \n",
       "shadowserver     0.49$\\pm$0.16  0.91$\\pm$0.01  0.90$\\pm$0.02  0.95$\\pm$0.01   \n",
       "driftnet         0.90$\\pm$0.02  0.88$\\pm$0.04  0.85$\\pm$0.05  0.97$\\pm$0.02   \n",
       "internetcensus   0.47$\\pm$0.13  0.64$\\pm$0.12  0.58$\\pm$0.14  0.90$\\pm$0.04   \n",
       "censys           0.60$\\pm$0.03  0.89$\\pm$0.03  0.88$\\pm$0.03  0.91$\\pm$0.02   \n",
       "rapid7           0.83$\\pm$0.10  0.84$\\pm$0.04  0.84$\\pm$0.02  0.69$\\pm$0.47   \n",
       "onyphe           0.05$\\pm$0.05  0.93$\\pm$0.03  0.93$\\pm$0.02  0.92$\\pm$0.01   \n",
       "netsystems       0.06$\\pm$0.05  0.97$\\pm$0.03  0.97$\\pm$0.02  0.80$\\pm$0.09   \n",
       "shodan           0.10$\\pm$0.14  0.81$\\pm$0.05  0.79$\\pm$0.06  0.76$\\pm$0.05   \n",
       "securitytrails   0.67$\\pm$0.04  0.98$\\pm$0.04  0.96$\\pm$0.05  0.97$\\pm$0.04   \n",
       "intrinsec        0.00$\\pm$0.00  0.32$\\pm$0.24  0.26$\\pm$0.19  0.75$\\pm$0.12   \n",
       "unk_bruteforcer  0.52$\\pm$0.01  0.62$\\pm$0.03  0.61$\\pm$0.03  0.61$\\pm$0.03   \n",
       "unk_spammer      0.19$\\pm$0.07  0.47$\\pm$0.06  0.46$\\pm$0.06  0.42$\\pm$0.08   \n",
       "unk_exploiter    0.06$\\pm$0.16  0.07$\\pm$0.20  0.07$\\pm$0.20  0.00$\\pm$0.00   \n",
       "macro avg        0.41$\\pm$0.04  0.74$\\pm$0.02  0.73$\\pm$0.03  0.75$\\pm$0.03   \n",
       "\n",
       "                     iGCNGRU_f  \n",
       "0                               \n",
       "mirai            0.98$\\pm$0.00  \n",
       "shadowserver     0.96$\\pm$0.01  \n",
       "driftnet         0.97$\\pm$0.02  \n",
       "internetcensus   0.92$\\pm$0.02  \n",
       "censys           0.92$\\pm$0.01  \n",
       "rapid7           0.97$\\pm$0.04  \n",
       "onyphe           0.91$\\pm$0.05  \n",
       "netsystems       0.95$\\pm$0.04  \n",
       "shodan           0.79$\\pm$0.06  \n",
       "securitytrails   0.98$\\pm$0.03  \n",
       "intrinsec        0.73$\\pm$0.13  \n",
       "unk_bruteforcer  0.63$\\pm$0.03  \n",
       "unk_spammer      0.47$\\pm$0.08  \n",
       "unk_exploiter    0.10$\\pm$0.21  \n",
       "macro avg        0.80$\\pm$0.02  "
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tqdm.notebook import tqdm_notebook as tqdm\n",
    "\n",
    "columns = ['iDarkVec', 'GCN', 'iGCN', 'GCNGRU', 'iGCNGRU', \n",
    "           'GCN_f', 'iGCN_f', 'GCNGRU_f', 'iGCNGRU_f']\n",
    "\n",
    "final_reps = []\n",
    "for col_name in columns:\n",
    "    # Manage names\n",
    "    if col_name == 'iDarkVec': \n",
    "        emb_path = f'../data/nlp_embeddings/'\n",
    "        fname = f'{col_name.lower()}_'\n",
    "    else: \n",
    "        emb_path = f'../data/gnn_embeddings/'\n",
    "        if '_f' in col_name: \n",
    "            fname = f'{col_name.lower()[:-2]}_features_'\n",
    "        else: \n",
    "            fname = f'{col_name.lower()}_'\n",
    "\n",
    "    reports = []\n",
    "    # For each testing snapshots\n",
    "    pbar = tqdm(total=len(range(21, 32)))\n",
    "    for day in range(21, 32):\n",
    "        pbar.set_description(desc=f'{col_name} -- Day {day}')\n",
    "        # Load active senders\n",
    "        active = pd.read_csv(f'../data/raw/raw_202112{day}.csv')\\\n",
    "                   .src_ip.unique()\n",
    "        \n",
    "        # Retrieve embeddings\n",
    "        embeddings = pd.read_csv(\n",
    "            f'{emb_path}/embeddings_{fname}202112{day}.csv', \n",
    "            index_col=[0]\n",
    "        )\n",
    "        \n",
    "        # Run classification and extract F1-Score\n",
    "        crep = classification_pipeline(embeddings, active, gt)\n",
    "        reports.append(pd.DataFrame(crep).T['f1-score'])\n",
    "        pbar.update(1)\n",
    "        \n",
    "    # Get mean and std of F1-Score\n",
    "    mean = pd.concat(reports, axis=1).mean(axis=1, skipna=True)\n",
    "    std = pd.concat(reports, axis=1).std(axis=1, skipna=True)\n",
    "    \n",
    "    # Manage results as a dataframe\n",
    "    idx = mean.index\n",
    "    items = [(idx[i], f'{x[0]:.2f}$\\pm${x[1]:.2f}') \n",
    "                         for i, x in enumerate(zip(mean,std))]\n",
    "    final_rep = pd.DataFrame(items).set_index(0).rename(columns={1:col_name})\n",
    "    final_reps.append(final_rep)\n",
    "\n",
    "# Concatenate everything and save\n",
    "df = pd.concat(final_reps, axis=1).reindex(indices)\n",
    "df.to_csv('../data/results/main_results.csv')\n",
    "\n",
    "df"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b26a1112-d4dc-481e-9573-29c90210c692",
   "metadata": {
    "tags": []
   },
   "source": [
    "## 2. Impact of History $H$"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5e007b56-592c-4fef-aeab-63eba8f39808",
   "metadata": {},
   "source": [
    "The following code performs the same classification pipeline of the main results, but only for the i-GCN-GRU model and for different values of the history parameter $H$."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "5fc5df8b-248e-4668-b1b6-555be5c854aa",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "af3f8f54e4604c0db03bd6bf24437e58",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/11 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "d6b26bdb6e8146ffb4dc4589ef24c5e4",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/11 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "7e83d7923c5a406e93588cd6e1bb3dd7",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/11 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "4f1d9343bd7c48519f23dfb113f6dc36",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/11 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "04ebbd0d54e54c4187dd89a3d5d4da9b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/11 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f9272d0fd8ff4ba1987e32c54d3cfb0f",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/11 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "44af2d9960f54460ac0fa4c09f6d7217",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/11 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "7e75932fc50a450ab0688131aa0bb655",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/11 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "db322de9aabd4287b45139e6f855b7bf",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/11 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>h</th>\n",
       "      <th>f1</th>\n",
       "      <th>f1min</th>\n",
       "      <th>f1max</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0.729475</td>\n",
       "      <td>0.707881</td>\n",
       "      <td>0.751070</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>0.756282</td>\n",
       "      <td>0.736731</td>\n",
       "      <td>0.775833</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>0.780522</td>\n",
       "      <td>0.763755</td>\n",
       "      <td>0.797289</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>0.791265</td>\n",
       "      <td>0.773972</td>\n",
       "      <td>0.808557</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>0.790726</td>\n",
       "      <td>0.775141</td>\n",
       "      <td>0.806310</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>5</td>\n",
       "      <td>0.796390</td>\n",
       "      <td>0.780878</td>\n",
       "      <td>0.811903</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>6</td>\n",
       "      <td>0.793773</td>\n",
       "      <td>0.779929</td>\n",
       "      <td>0.807617</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>7</td>\n",
       "      <td>0.797479</td>\n",
       "      <td>0.781963</td>\n",
       "      <td>0.812995</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>8</td>\n",
       "      <td>0.795375</td>\n",
       "      <td>0.776371</td>\n",
       "      <td>0.814379</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   h        f1     f1min     f1max\n",
       "0  0  0.729475  0.707881  0.751070\n",
       "1  1  0.756282  0.736731  0.775833\n",
       "2  2  0.780522  0.763755  0.797289\n",
       "3  3  0.791265  0.773972  0.808557\n",
       "4  4  0.790726  0.775141  0.806310\n",
       "5  5  0.796390  0.780878  0.811903\n",
       "6  6  0.793773  0.779929  0.807617\n",
       "7  7  0.797479  0.781963  0.812995\n",
       "8  8  0.795375  0.776371  0.814379"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tqdm.notebook import tqdm_notebook as tqdm\n",
    "\n",
    "reports = []\n",
    "emb_path = f'../data/gnn_embeddings/'\n",
    "for h in [0, 1, 2, 3, 4, 5, 6, 7, 8]:\n",
    "    if h!=5:\n",
    "        fname = f'igcngru_features_H{h}_'\n",
    "    else:\n",
    "        fname = f'igcngru_features_'\n",
    "        \n",
    "    pbar = tqdm(total=len(range(21, 32)))\n",
    "    # For each testing snapshots\n",
    "    for day in range(21, 32):\n",
    "        pbar.set_description(desc=f'H {h} -- Day {day}')\n",
    "        # Load active senders\n",
    "        active = pd.read_csv(f'../data/raw/raw_202112{day}.csv')\\\n",
    "                   .src_ip.unique()\n",
    "        \n",
    "        # Retrieve embeddings\n",
    "        embeddings = pd.read_csv(\n",
    "            f'{emb_path}/embeddings_{fname}202112{day}.csv', \n",
    "            index_col=[0]\n",
    "        )\n",
    "        \n",
    "        # Run classification and extract F1-Score\n",
    "        crep = classification_pipeline(embeddings, active, gt)\n",
    "        reports.append((h, crep['macro avg']['f1-score']))\n",
    "        \n",
    "        pbar.update(1)\n",
    "\n",
    "# Manage results in dataframe\n",
    "df = pd.DataFrame(reports, columns=['h', 'f1'])\n",
    "\n",
    "# Get avg. and std. of F1-Score\n",
    "df = df.groupby('h')['f1'].agg(['mean', 'std']).reset_index()\n",
    "df.rename(columns={'mean': 'f1', 'std': 'stdf1'}, inplace=True)\n",
    "\n",
    "# Get min and max\n",
    "df['f1min'] = df['f1'] - df['stdf1']\n",
    "df['f1max'] = df['f1'] + df['stdf1']\n",
    "df = df.drop(columns=['stdf1'])\n",
    "\n",
    "# Save\n",
    "df.to_csv('../data/results/impact_of_history.csv', index=False)\n",
    "\n",
    "df"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "01847a55-2741-4a93-a469-2a17a57cf2b3",
   "metadata": {
    "tags": []
   },
   "source": [
    "## 3. Impact of Training Epochs"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5a71a52a-8e42-4b8d-9679-99400277e033",
   "metadata": {},
   "source": [
    "The following code performs the same classification pipeline of the main results, but only for the i-GCN-GRU model and for different values of the training epochs used in the incremental training procedure."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "27fcded3-741a-45f4-9ada-9857feb060d3",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a2cf33a03b9f404b9eee83b297d966e0",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/11 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f9c1130b9e254f5892ea346236903e4b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/11 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "4da87cf4c2674fec8398ae85d15ba649",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/11 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "81e1c0493946491689c4a30a032baf7a",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/11 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "93a6644e695a4e1ba02182740f9473a5",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/11 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2452a529195e4fd2b12c0e043ea6b475",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/11 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "fc82e5037a5747618cdf2c3c3b9cfe72",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/11 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>epoch</th>\n",
       "      <th>f1</th>\n",
       "      <th>f1min</th>\n",
       "      <th>f1max</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>0.796390</td>\n",
       "      <td>0.780878</td>\n",
       "      <td>0.811903</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>5</td>\n",
       "      <td>0.786023</td>\n",
       "      <td>0.774541</td>\n",
       "      <td>0.797506</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>10</td>\n",
       "      <td>0.794721</td>\n",
       "      <td>0.780886</td>\n",
       "      <td>0.808556</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>15</td>\n",
       "      <td>0.788198</td>\n",
       "      <td>0.774631</td>\n",
       "      <td>0.801766</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>20</td>\n",
       "      <td>0.778478</td>\n",
       "      <td>0.762163</td>\n",
       "      <td>0.794793</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>25</td>\n",
       "      <td>0.771929</td>\n",
       "      <td>0.759982</td>\n",
       "      <td>0.783876</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>30</td>\n",
       "      <td>0.759843</td>\n",
       "      <td>0.747208</td>\n",
       "      <td>0.772478</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   epoch        f1     f1min     f1max\n",
       "0      1  0.796390  0.780878  0.811903\n",
       "1      5  0.786023  0.774541  0.797506\n",
       "2     10  0.794721  0.780886  0.808556\n",
       "3     15  0.788198  0.774631  0.801766\n",
       "4     20  0.778478  0.762163  0.794793\n",
       "5     25  0.771929  0.759982  0.783876\n",
       "6     30  0.759843  0.747208  0.772478"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tqdm.notebook import tqdm_notebook as tqdm\n",
    "\n",
    "reports = []\n",
    "emb_path = f'../data/gnn_embeddings/'\n",
    "for epoch in [1, 5, 10, 15, 20, 25, 30]:\n",
    "    if epoch!=1:\n",
    "        fname = f'igcngru_features_ee{epoch}_'\n",
    "    else:\n",
    "        fname = f'igcngru_features_'\n",
    "        \n",
    "    pbar = tqdm(total=len(range(21, 32)))\n",
    "    # For each testing snapshots\n",
    "    for day in range(21, 32):\n",
    "        pbar.set_description(desc=f'Epochs {epoch} -- Day {day}')\n",
    "        # Load active senders\n",
    "        active = pd.read_csv(f'../data/raw/raw_202112{day}.csv')\\\n",
    "                   .src_ip.unique()\n",
    "        \n",
    "        # Retrieve embeddings\n",
    "        embeddings = pd.read_csv(\n",
    "            f'{emb_path}/embeddings_{fname}202112{day}.csv', \n",
    "            index_col=[0]\n",
    "        )\n",
    "        \n",
    "        # Run classification and extract F1-Score\n",
    "        crep = classification_pipeline(embeddings, active, gt)\n",
    "        reports.append((epoch, crep['macro avg']['f1-score']))\n",
    "        \n",
    "        pbar.update(1)\n",
    "\n",
    "# Manage results in dataframe\n",
    "df = pd.DataFrame(reports, columns=['epoch', 'f1'])\n",
    "\n",
    "# Get avg. and std. of F1-Score\n",
    "df = df.groupby('epoch')['f1'].agg(['mean', 'std']).reset_index()\n",
    "df.rename(columns={'mean': 'f1', 'std': 'stdf1'}, inplace=True)\n",
    "\n",
    "# Get min and max\n",
    "df['f1min'] = df['f1'] - df['stdf1']\n",
    "df['f1max'] = df['f1'] + df['stdf1']\n",
    "df = df.drop(columns=['stdf1'])\n",
    "\n",
    "# Save\n",
    "df.to_csv('../data/results/impact_of_epochs.csv', index=False)\n",
    "\n",
    "df"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8be0d376-4449-4ca3-8487-367b8af18988",
   "metadata": {
    "tags": []
   },
   "source": [
    "## 4. Visualization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "a7598dfe-da71-4014-8b42-5fb7b54de0df",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 640x480 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 350x350 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import fastplot\n",
    "import pandas as pd\n",
    "\n",
    "df = pd.read_csv('../data/results/impact_of_epochs.csv')\n",
    "\n",
    "def plotEpochs(plt):\n",
    "    plt.plot(df.epoch, df.f1, marker='o', label='i-GCN-GRU', \n",
    "             color='purple', alpha=1, linestyle='-')\n",
    "    plt.fill_between(df.epoch, df.f1min, df.f1max, \n",
    "             color='purple', alpha=.2)\n",
    "    \n",
    "    \n",
    "    plt.xlabel('Training epochs')\n",
    "    plt.ylabel('Avg. F1-Score')\n",
    "    plt.grid()\n",
    "    plt.xticks([1, 5, 10, 15, 20, 25, 30])\n",
    "    plt.xlim(1, 30)\n",
    "    \n",
    "%matplotlib inline\n",
    "plot = fastplot.plot(None,  None, mode = 'callback', \n",
    "                     callback = lambda plt: plotEpochs(plt),\n",
    "                     figsize=(3.5, 3.5), fontsize=14, style='latex')\n",
    "plot.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "6d22585b-13aa-4f3a-ac9b-ff88c5b9abaa",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 640x480 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 350x350 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import fastplot\n",
    "import pandas as pd\n",
    "\n",
    "df = pd.read_csv('../data/results/impact_of_history.csv')\n",
    "\n",
    "def plotHistory(plt):\n",
    "    plt.plot(df.h, df.f1, marker='o', label='i-GCN-GRU', \n",
    "             color='purple', alpha=1, linestyle='-')\n",
    "    plt.fill_between(df.h, df.f1min, df.f1max, \n",
    "             color='purple', alpha=.2)\n",
    "    \n",
    "    \n",
    "    plt.xlabel('History length $H$')\n",
    "    plt.ylabel('Avg. F1-Score')\n",
    "    plt.grid()\n",
    "    plt.xticks([0, 2, 4, 6, 8])\n",
    "    plt.xlim(0, 8)\n",
    "    \n",
    "%matplotlib inline\n",
    "plot = fastplot.plot(None,  None, mode = 'callback', \n",
    "                     callback = lambda plt: plotHistory(plt),\n",
    "                     figsize=(3.5, 3.5), fontsize=14, style='latex')\n",
    "plot.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.9.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}