Look at the Default dataset from the ISLR package. It
contains information on defaults of credit card payments. Fit some
models to predict which payments will default.
Try fitting generalised linear models and generalised additive models. Compare the results of the different models in prediction on test data, and the model diagnostics.