Skip to content

Latest commit

 

History

History
33 lines (27 loc) · 1.53 KB

README.md

File metadata and controls

33 lines (27 loc) · 1.53 KB

Fraud Detection Using Machine Learning

Overview

This project explores machine learning techniques to detect fraudulent transactions in a highly imbalanced dataset. The dataset contains anonymized transaction details with the goal of identifying fraudulent transactions (Class 1).

Dataset

The dataset consists of:

  • Time: Time elapsed since the first transaction.
  • V1-V28: Anonymized features.
  • Amount: Transaction amount.
  • Class: Target variable indicating fraud (1) or non-fraud (0).

Class Distribution:

  • Fraudulent Transactions: 492 instances (0.17%)
  • Non-Fraudulent Transactions: 284,315 instances (99.83%)

Key Models and Results

Model ROC AUC Precision Recall Notes
Logistic Regression 0.9559 0.83 0.65 Simple and interpretable.
Decision Tree 0.8875 0.75 0.78 Prone to overfitting.
Random Forest 0.9578 0.94 0.82 Robust and feature-rich.
SVM 0.9646 0.96 0.67 Effective but complex.
XGBoost 0.9711 0.89 0.83 Best performance overall.

Files

  • analysis/DSC478_Final_Project.ipynb: Jupyter notebook containing the analysis and code.
  • report/Finalreport-PML.pdf: Comprehensive project report.

How to Use

  1. Clone the repository: 
    git clone https://github.com/tejas-1911/Fraud-Detection-Using-Machine-Learning.git