You are tasked with building a predictive model for a bank to determine whether a customer will default on a loan. The dataset includes features such as age, income, credit score, employment status, and loan amount. The dataset is significantly imbalanced, with only $5\%$ of the instances representing defaults.

How would you handle the class imbalance in the dataset?

Which ensemble learning techniques would you apply, and why?

How would you use grid search to optimize your model parameters?

How would you estimate the confidence measures of your predictions?

Explain how you would compute and interpret the feature importance for this model.