This repository contains code to download a fraud detection dataset, train a model, and run an application to predict fraud based on transactions.

For Unix-based systems (Linux, macOS):

sh download.sh

For Windows systems, you can manually download the dataset from Kaggle and place it in the fraud_detection directory.

Run the train_model.py script to train the fraud detection model.

python train_model.py

This script will:

• Load the dataset from fraudTrain.csv.
• Split the data into training and testing sets.
• Train a RandomForest classifier.
• Save the trained model to fraud_model.pkl.

Run the app.py script to start the Flask application.

python app.py

This script will:

• Load the trained model.
• Start a Flask web server.
• Provide an endpoint /predict for making fraud predictions.

Method: POST

Request Body: JSON object containing the following fields:

• cc_num
• amt
• zip
• lat
• long
• city_pop
• unix_time
• merch_lat
• merch_long

Response:

• prediction: 1 if the transaction is predicted to be fraud, 0 otherwise.

Example request using curl:

curl -X POST http://127.0.0.1:5000/predict -H "Content-Type: application/json" -d '{
  "cc_num": "1234567890123456",
  "amt": 123.45,
  "zip": "12345",
  "lat": 40.7128,
  "long": -74.0060,
  "city_pop": 100000,
  "unix_time": 1615158000,
  "merch_lat": 40.730610,
  "merch_long": -73.935242
}'

Response Example:

{
  "prediction": 1
}

This project is licensed under the MIT License. See the LICENSE file for details.