Credit Risk Scoring System

A comprehensive credit risk analysis and scoring platform built with Python 3.12 and PyQt6. This system enables financial institutions to assess customer default risk using advanced machine learning models, a user-friendly interface, and robust data management.

Table of Contents

• Overview
• Features
• Technology Stack
• Project Structure
• Prerequisites
• Installation
• Database Setup
• Running the Application
• Usage Guide
• Team Members
• Contributing
• License

Overview

The Credit Risk Scoring System is a full-stack application designed to automate and enhance the process of credit risk assessment. It provides:

• A modern desktop interface for data entry, prediction, and reporting.
• Machine learning models (XGBoost, LightGBM, Logistic Regression) for risk scoring.
• Admin and user roles with tailored access and management features.
• Data visualization and model comparison tools.

Features

• Customer data management and search
• Credit risk prediction using multiple ML models
• Model comparison and performance metrics (ROC-AUC, PR-AUC, Accuracy, etc.)
• Admin dashboard for model management and user administration
• Multi-currency support (VND/NT$)
• Audit logs and prediction history
• AI Assistant (Gemini integration, optional)

Technology Stack

• Python 3.12 (recommended)
• PyQt6 for GUI
• scikit-learn, xgboost, lightgbm for ML
• MySQL for database
• pandas, numpy, matplotlib for data processing and visualization

Project Structure

MLBA_FinalProject/
├── main.py                  # Main entry point
├── ui/                      # UI components (PyQt6)
├── services/                # Business logic and database services
├── ml/                      # Machine learning scripts and models
├── models/                  # Data models
├── outputs/                 # Model files, evaluation results, charts
├── database/credit_risk_db/ # SQL schema files (import here!)
├── config/                  # Configuration files
├── requirements.txt         # Python dependencies
└── README.md                # This file

Prerequisites

• Python 3.12 (strongly recommended)
• MySQL server (tested with MySQL 8+)
• pip (Python package manager)

Database Setup

1. Start MySQL Server and create a database (e.g., credit_risk_db).
2. Import Data Schema
   • Open your MySQL client (Workbench, DBeaver, or command line).
   • Import all .sql files from the folder:

     database/credit_risk_db/
     

   • This will create all necessary tables: users, customers, predictions_log, model_registry, etc.
3. (Optional) Update database credentials
   • Edit config/database_config.py if your MySQL username/password is different.

Running the Application

1. Train Machine Learning Models (first time only)

   python ml/train_models.py

   This will generate model files and evaluation data in outputs/models/ and outputs/evaluation/.

2. Start the Application

   python main.py
   # or
   py -3.12 main.py

Usage Guide

• On launch, the login screen will appear. Use the following demo accounts:

  Admin:

  • Username: ilovetranduythanh2
  • Password: 10diem10diem

  User:

  • Username: ilovetranduythanh1
  • Password: 10diem10diem

• Admin can manage users, train models, view all predictions, and access system settings.

• User can input customer data, make predictions, and view their own prediction history.

• Use the "Model Management" tab (Admin) to train or activate new models.

• All data is stored in the MySQL database.

Contributing

1. Fork the repository.
2. Create a feature branch (git checkout -b feature/YourFeature).
3. Commit your changes (git commit -m 'Add YourFeature').
4. Push to your branch (git push origin feature/YourFeature).
5. Open a Pull Request.

License

This project is for educational purposes only. All rights reserved by the development team.