Credit Card Fraud Detection Pipeline

This repository contains an end-to-end machine learning pipeline to detect fraudulent credit card transactions.

Project Structure

• eda.py: Performs Exploratory Data Analysis and generates visualizations.
• preprocess.py: Scales data using RobustScaler and handles imbalance with SMOTE.
• train.py: Trains Logistic Regression and Random Forest models.
• evaluate.py: Generates performance metrics (AUPRC, Recall, Precision) and plots.
• predict.py: Provides an interface for making predictions on new transaction data.

Performance Summary

Model	AUPRC	Precision (Fraud)	Recall (Fraud)
Logistic Regression	0.734	0.06	0.90
Random Forest	0.796	0.51	0.84

Quick Start

1. Explore Data: python eda.py
2. Preprocess State: python preprocess.py
3. Train Models: python train.py
4. Evaluate Performance: python evaluate.py
5. Run Inference:

from predict import predict_fraud
result = predict_fraud(transaction_features)
print(result)

Visualizations

Check the following generated files in the root directory:

• class_distribution.png
• precision_recall_curve.png
• confusion_matrix_random_forest.png

# Fraud-Detection