This project was part of a requirement for MSBA 6420, Predictive Analytics. I worked with Callie Page, Max Traub, and Cathy Zheng. The data comes from a 2018 closed Kaggle competition.
Home Credit, an international consumer finance provider, helps people with limited credit history get loans. Since the company tailors to a non traditional market, they use alternative data to predict repayment ability. This includes telco and transactional information, as well as previous application data. While they already use machine learning models to make predictions, they improve their predictive capabilities with a Light GBM model. We build the best performing model in class history, that performs very well compared to other Kagglers as well.