Credit Card Default Prediction Using Machine Learning
A supervised-learning study on an imbalanced 30,000-client dataset (~22% default), comparing five classifiers with a focus on minority-class performance rather than misleading accuracy. Course project at École polytechnique.