Cost-Sensitive Predictive Modeling for SBA Loan Approval 🏦📊

This project builds a cost-sensitive predictive model for SBA loan approval decisions using R and Python. The model optimizes loan classification by minimizing misclassification costs and maximizing net profit.

📌 Features

• Data Preprocessing: Cleaned and prepared SBA loan data for analysis.
• Logistic Regression & Regularized Models: Implemented Lasso, Ridge, and Elastic Net in R.
• Hyperparameter Tuning: Used cross-validation to optimize model performance.
• Advanced Predictive Analytics: Applied neural networks, decision trees, random forests, and discriminant analysis in Python.
• Performance Evaluation: Assessed models with confusion matrices, gains/lift charts, and net profit analysis.
• Actionable Insights: Identified the most profitable loan approval strategy for data-driven decision-making.

🔧 Technologies Used

• R: glmnet, caret, tidyverse
• Python: scikit-learn, pandas, numpy, matplotlib