Credit Risk & Loan Default Prediction

Project Overview

Developed supervised machine learning models to predict payday loan default risk using structured financial and behavioral data
Performed data preprocessing and feature engineering to improve signal quality and model generalization
Evaluated multiple classification algorithms, identifying LightGBM and CatBoost as top-performing models compared to baseline approaches
Built a feedforward neural network (FFN) to benchmark deep learning performance against traditional ML models
Assessed model performance using appropriate classification metrics to guide model selection and tuning
Demonstrated real-world risk modeling skills applicable to finance, lending, and decision-support systems

Machine learning models for credit risk assessment predicting payday loan defaults using LightGBM, CatBoost, and feedforward neural networks.

Python Machine Learning Classification LightGBM CatBoost Neural Networks Feature Engineering Model Evaluation Applied Data Science