A comprehensive end-to-end data science project to analyze and predict credit risk using the Lending Club loan dataset.

View Full HTML Report

This project performs an in-depth credit risk analysis on the Lending Club dataset sourced from Kaggle. The goal is to identify key drivers of loan default and build a predictive model that classifies borrowers as high or low credit risk.

The project covers the full data science pipeline — from raw data cleaning, thorough exploratory data analysis and statistical testing to model building and deployment as an interactive web application.

• Identified the most significant predictors of credit default using statistical techniques including KS Test, Mutual Information, and Cramér's V.
• Borrowers with higher debt-to-income (DTI) ratios and lower annual income showed significantly higher default rates.
• Loan grade and sub-grade assigned by Lending Club proved to be strong indicators of credit risk.
• The final predictive model achieved strong performance metrics, with features engineered from raw financial variables contributing meaningfully to accuracy.
• Findings were documented in a structured HTML report covering domain context, methodology, and actionable insights.

• Source: Lending Club Loan Dataset — Kaggle
• Size: ~1.1 GB (not included in this repo due to GitHub file size limits)
• Description: Contains historical loan data from Lending Club including borrower information, loan attributes, and repayment status.

git clone https://github.com/mahatarc/Lending_club_credit_risk.git
cd Lending_club_credit_risk

pip install -r requirements.txt

jupyter notebook Lending_club_credit_risk_analysis.ipynb

streamlit run app.py

Then open your browser at http://localhost:8501

uvicorn main:app --reload

Then open your browser at http://localhost:8000
Interactive API docs available at http://localhost:8000/docs

The credit risk prediction model is deployed as an interactive web app built with Streamlit and FastAPI.