EnYa32's picture
Update README.md
c627435 verified
|
Raw History Blame Contribute Delete
1.51 kB
metadata
title: FaudDetectionClassification
emoji: πŸ’³
colorFrom: indigo
colorTo: pink
sdk: docker
app_port: 8501
tags:
  - streamlit
pinned: false
short_description: Fraud detection for online payments using a Random Forest
license: mit

πŸ’³ Payment Fraud Detection

This project demonstrates how machine learning can be used to detect fraudulent online payment transactions.

The Streamlit app allows users to enter transaction details and receive a prediction:

  • 0 β†’ Not Fraud
  • 1 β†’ Fraud

πŸš€ Application Overview

The app predicts fraud based on transaction-related features such as:

  • Account age
  • Number of purchased items
  • Transaction time
  • Payment method and payment method age

The goal is to accurately detect fraudulent transactions while minimizing false alarms.


πŸ€– Model

  • Algorithm: Random Forest Classifier
  • Problem Type: Binary Classification (Fraud vs. Not Fraud)
  • Class Imbalance Handling: SMOTE (Synthetic Minority Oversampling Technique)
  • Evaluation Metrics: Accuracy, Precision, Recall, F1-score, ROC-AUC

Among all tested models, Random Forest achieved the best performance and was selected as the final model.


πŸ“Š Model Performance (Summary)

  • Excellent separation between fraud and non-fraud transactions
  • High precision and recall for fraud detection
  • Robust performance on imbalanced data

πŸ›  How to Run Locally

pip install -r requirements.txt
streamlit run app.py