Home Loan Approval Prediction

About the Project

This classification project predicts whether a home loan application will be approved or rejected based on applicant features like income, credit history, loan amount, and property area. A Support Vector Machine (SVM) was selected after comparing multiple algorithms including Logistic Regression and Decision Trees. Extensive preprocessing was applied including missing value imputation, label encoding, and feature scaling. The final model achieved strong accuracy and recall on the test set, making it viable for real-world use as a pre-screening tool.

⚡ Challenges Faced

• Significant missing data across multiple columns required careful imputation strategies — using mode for categorical and median for numerical features.
• Credit history was highly correlated with the target variable, raising concerns about data leakage that needed to be carefully managed.
• Selecting the right SVM kernel (RBF vs linear) and tuning the C and gamma parameters required extensive grid search cross-validation.

🚀 Future Plans & Improvements

• Build a REST API using FastAPI to serve the trained model for real-time predictions.
• Add a React frontend where applicants can input their details and receive instant pre-screening feedback.
• Extend the model with fairness-aware constraints to ensure unbiased approval predictions across demographic groups.