How anuragaryanyt/upifrauddetection works
Project type and purpose
• This is a full-stack web application for UPI fraud detection.
• Purpose: ingest UPI transaction data, train/evaluate ML models, and provide an interactive dashboard for real-time detection and visualization of fraud.
File structure
• Top-level files:
• app.py – Flask web server & routing
• requirements.txt – Python dependencies
• README.md – project overview
• Directories:
• dataset/ – raw CSV data used for training & evaluation
• filesuse/ – persisted trained model (project_model1.h5)
• src/ – Jupyter notebooks & scripts for data processing, feature engineering, model training
• static/ – CSS, JS, images for the front end
• templates/ – HTML templates rendered by Flask
• Patterns:
• MVC-like separation – templates (View), app.py controllers (Controller), model files & dataset (Model)
• Machine learning pipeline separated from runtime inference
README for additional insights
• ML workflow: data cleaning → feature engineering → model training → prediction
• Interactive dashboard: upload data, view charts, alerts on fraud
• Tech stack: Python, Flask, scikit-learn (or TF/Keras for .h5), Jupyter, HTML/CSS/JS
Detail
• Every folder’s role: static assets vs. templates vs. notebooks
• External dependencies (requirements.txt)
• Where model files live and when they are loaded
• Both real-time inference flow and offline training flow