Credit Risk Predictor

Overview

This application predicts credit risk using machine learning algorithms applied to German credit data. It features a comprehensive preprocessing pipeline, multiple trained models including Random Forest, Logistic Regression, SVC, and XGBoost, with a stacking ensemble for optimal performance.

Features

• Interactive Web Interface: User-friendly Streamlit application for real-time credit risk predictions
• Advanced Preprocessing: Feature engineering, outlier handling, and data imputation techniques
• Ensemble Modeling: Stacking ensemble approach combining multiple ML algorithms
• Model Interpretation: SHAP analysis for explainable AI
• Performance Metrics: Confusion matrices and ROC curves for transparent model evaluation

Tech Stack

• Python: Core programming language
• Scikit-learn: Machine learning algorithms and preprocessing
• XGBoost: Gradient boosting implementation
• SMOTE: Handling class imbalance
• Streamlit: Web application framework
• Pandas/NumPy: Data manipulation
• Matplotlib/Plotly: Data visualization
• SHAP: Model interpretation

Project Structure

.
├── creditrisk.py              # Preprocessing and feature engineering
├── train_models.py            # Model training script
├── shap_analysis.py           # Model interpretation with SHAP
├── performance_plots.py       # Generate performance visualizations
├── credit_risk_streamlitapp.py # Web application
├── data/
│   ├── german_credit_data.csv  # Original dataset
│   └── preprocessed_credit_data.csv # Preprocessed data
├── models/                    # Saved models and encoders
└── plots/                     # Performance and interpretation plots

Installation

# Clone the repository
git clone https://github.com/Kpreya/Credit_Risk_Predictor_App.git
cd Credit_Risk_Predictor_App

# Install dependencies
pip install -r requirements.txt

Usage

1. Data Preprocessing

python creditrisk.py

This script loads the German credit dataset, performs cleaning, feature engineering, and preprocessing tasks, and saves the artifacts for model training.

2. Model Training

python train_models.py

Trains multiple machine learning models and saves them for later use.

3. Model Analysis

python shap_analysis.py
python performance_plots.py

Generates model interpretation plots and performance metrics.

4. Running the Web App

streamlit run credit_risk_streamlitapp.py

Launches the interactive web application for credit risk prediction.

Model Performance

The application employs a stacking ensemble of multiple base models to achieve optimal prediction performance. Model evaluation metrics include:

• Confusion matrix
• ROC curve
• SHAP feature importance
• Cross-validation scores

Feature Importance

Key factors in credit risk prediction include:

• Financial security (saving level relative to credit amount)
• Checking account status
• Credit amount and duration
• Savings account level
• Purpose of the loan
• Age and employment status

Future Enhancements

• Additional model algorithms for performance comparison
• Deep learning approaches for complex pattern recognition
• API deployment for integration with banking systems
• Enhanced visualization options for better interpretability

License

MIT

Author

Developed by Krishnopreya C. Contact :krish6.ch@gmail.com

Acknowledgments

• German Credit Data from UCI Machine Learning Repository
• SHAP library for model interpretation
• Streamlit for web application framework