Click-Through Rate Prediction with XGBoost and Flask

This project demonstrates how to deploy a machine learning model using Flask. Specifically, it includes an XGBoost classifier for predicting click-through rates from advertising data. The model is integrated into a Flask web application for real-time predictions.

Features

• XGBoost Model: Trained on advertising data to predict click-through rates.
• Flask Web Application: Provides a RESTful API endpoint for making predictions.

Requirements

• Python 3.x
• Flask
• pandas
• scikit-learn
• XGBoost
• cloudpickle
• joblib

Installation

1. Clone the repository:

   git clone https://github.com/your-username/your-repository.git
   cd your-repository
   

2. Install dependencies:

   pip install -r requirements.txt
   

##Usage 1.Run the Flask Application:

python app.py

2. Make Predictions: Send a POST request to http://localhost:5000/predict with JSON data containing the features for prediction. Example:

{
  "Daily Time Spent on Site": value1,
  "Age": value2,
  "Area Income": value3,
  "Daily Internet Usage": value4,
  "Ad Topic Line": "value5",
  "City": "value6",
  "Gender": "value7",
  "Country": "value8",
  "Timestamp": "value9"
}

3. Response: The server will respond with a JSON object containing the prediction:

{
  "prediction": 1 # 0 or 1
}

Note: Make sure to change directories used in click-through-rate-prediction.ipynb according to your local directories if you want to run it.

app.py: Flask application for serving predictions.

preprocessor.pkl: Preprocessor serialized using cloudpickle.

clickthroughmodel.pkl: XGBoost model serialized using joblib.

requirements.txt: List of Python dependencies.

data.zip: contains the dataset this model was trained on.

link: https://www.kaggle.com/datasets/swekerr/click-through-rate-prediction