Skip to content

README.md

Latest commit

 

History

History
210 lines (133 loc) · 5.03 KB

README.md

File metadata and controls

210 lines (133 loc) · 5.03 KB

💰 Personal Loan Default Prediction - Streamlit Application

Live Demo: Click here to try the app 🚀

Animated Header

🧹 Problem Statement

Financial institutions face increasing challenges in accurately identifying applicants who might default on personal loans. Misclassification can lead to bad debts, regulatory issues, and loss of trust. A reliable prediction system can help mitigate risk, enhance operational efficiency, and drive smarter lending decisions.

🎯 Goal of the Project

  • ✅ Predict the probability of loan default based on customer attributes
  • ✅ Provide a production-ready, interactive Streamlit app for banks and analysts
  • ✅ Explain predictions with SHAP visualizations
  • ✅ Export reports to share insights with decision-makers

🛠️ Flow Diagram

graph TD
    A[User Inputs Loan Details via Form] --> B[Data Preprocessing]
    B[Data Preprocessing] -->C[ML model]
    C[ML model] -->E[Predict] 
    E[Predict] --> F[Recommendation]
Loading

📂 Folder Structure

Folder Structure

🛠️ Tech Stack

Layer Technology
Frontend Streamlit
Backend Python
ML Models Logistic Regression, Random Forest, XGBoost
Visualization Plotly, Matplotlib, SHAP
Export Tools PDFKit, Streamlit Export

🧠 Code Walkthrough

# Load the model
model = pickle.load(open("model/model.pkl", "rb"))

# Preprocess user inputs
def preprocess(data):
    ...
    return processed_input

# Make prediction
prediction = model.predict(processed_input)

# Show results
st.success(f"Prediction: {'Approved' if prediction==0 else 'Rejected'}")

Automate EDA REPORT

🞼 Application UI & Screenshots

🔘 Home Page

Home Page

🔢 Prediction Form

Prediction Result

🔢 Prediction Result

Prediction Result

📉 ROC AUC Curve

ROC AUC CURVE

🎟️ Precision_Recall Curve

Precision_Recall_Curve

🎟️ Loan KPI

KPI

🎟️ 3D SCATTER

KPI

✨ Key Features

  • 🔹 User-friendly UI built in Streamlit
  • 🔹 Real-time loan eligibility prediction
  • 🔹 Support for multiple ML models
  • 🔹 SHAP visualizations for explainability
  • 🔹 Exportable PDF reports
  • 🔹 Clean, minimal, responsive design

🔮 Scientific Innovation

What makes this project stand out:

  • 🧠 Explainable AI (XAI): Integrates SHAP values to explain each prediction
  • 🧬 Bias Check Module: Identify model fairness issues across genders/ages
  • ✨ Risk Interpretation Layer: Converts numeric predictions into easy-to-understand advice
  • 🚀 Model Comparator: Visual comparison between Logistic, Random Forest & XGBoost

🚀 Future Enhancements

  • 🚀 Aadhaar/PAN Verification via Gemini API
  • 🎮 Voice-based AI Assistant for customer service
  • 🤖 Chatbot integration for insights
  • 📊 Historical dashboard for tracking applicant trends
  • 👨‍📈 Credit score simulator based on customer profile edits

🚧 How to Run Locally

# Step 1: Clone the repo
git clone https://github.com/yourusername/personal-loan-prediction.git
cd personal-loan-prediction

# Step 2: Install dependencies
pip install -r requirements.txt

# Step 3: Run the application
streamlit run app.py

🧳 Collaboration & Contribution

We'd love your help! You can:

  • ✨ Improve UI/UX
  • 🧠 Optimize the model pipeline
  • 📉 Add new data sources or APIs
  • 🤝 Translate into local languages

How to Contribute

# Fork the repository
# Create your feature branch
git checkout -b feature/new-awesome-feature

# Commit and push your changes
git commit -m "Added awesome feature"
git push origin feature/new-awesome-feature

# Submit a Pull Request

📢 Contact Me

Platform Link
💼 LinkedIn Ankit Yadav
💻 GitHub Ankit Yadav
📧 Email ankitofficial151@gmail.com

📄 License

This project is licensed under the MIT License. Feel free to fork, adapt, and build upon it for your use case!

💎 Pro Tip

Use this architecture as a blueprint for any classification-based real-time prediction system with SHAP explainability, PDF reporting, and clean UI design.

🔹 Tags

#LoanPrediction #StreamlitApp #ExplainableAI #SHAP #FinanceAI #BankingML #CreditScoring