Loan Analysis & Prediction Project

Built a complete loan analysis system from raw data to working web app. Covers everything from SQL data prep to deployed machine learning models.

What I Did

1. SQL Work: Combined 3 messy datasets, created new features, cleaned everything up
2. Tableau Dashboards: Made charts showing loan trends and borrower patterns
3. Python Models: Built models to predict loan amounts and spot risky loans
4. Web App: Deployed working app so people can actually use the models

The Models

• Simple Linear Regression: Basic model using just loan amount
• Multiple Linear Regression: Uses loan amount, charges, and duration - gets 99.7% accuracy
• Logistic Regression: Predicts if loans will be good or bad - had to fix class imbalance issues

Results That Actually Matter

Amount Prediction: Works great - can predict what someone will owe with 99.7% accuracy.

Risk Classification: Started terrible (missed 99% of bad loans), got much better after balancing the data. Now catches over half the risky loans instead of basically none.

What's In Here

• notebooks/: All the Python analysis and model building
• sql/: Queries I used to prep and combine the data
• models/: The actual trained models
• app/: Web app for making predictions
• tableau/: Dashboards and visualizations

Tools Used

• SQL for data engineering
• Tableau for business dashboards
• Python for machine learning
• Streamlit for the web app

Real working system that goes from messy data to predictions you can actually use.