R Studio Projects

Predicting Loan Repayment Outcomes Using Machine Learning Techniques

M.A in Applied Economics Program @ UNC Greensboro
Project Link
In this project, I applied four machine learning models to predict the repayment status of loan applicants—specifically, whether they would fully repay the loan or be classified as “charged off.” This classification helped assess each applicant’s eligibility for loan approval based on predicted financial behavior. I used a dataset containing 17 variables and split it into 70% for training and 30% for testing the models. To evaluate model performance, I generated a confusion matrix to analyze prediction accuracy.

The models I implemented included Logistic Regression, XGBoost, Over Sampling to address class imbalance, and a Decision Tree.