Credit Card Fraud Detection

Project Overview

This project implements a comprehensive machine learning pipeline to detect fraudulent credit card transactions. Given the highly imbalanced nature of fraud datasets, the project explores various advanced techniques—including Synthetic Minority Over-sampling Technique (SMOTE) and Deep Learning—to maximize the detection of fraudulent activities while minimizing false alarms.

Features

• Data Preprocessing: Implements both StandardScaler for feature standardization and MinMaxScaler for normalization.
• Imbalance Handling: Uses SMOTE with a sampling strategy of 0.4 to synthetically balance the dataset.
• Deep Learning Model: A Sequential Neural Network built with TensorFlow/Keras featuring BatchNormalization, Dropout, and EarlyStopping.
• Multi-Model Comparison: Explores Support Vector Machines (SVM), Logistic Regression, and Anomaly Detection (Isolation Forest).

Installation

pip install -r requirements.txt

Dataset

The project uses the Credit Card Fraud Detection dataset from Kaggle. The notebook automatically handles the download using kagglehub.

Usage

Open the Jupyter Notebook and run all cells. The first cell will handle the data download.