With an accuracy score of 82.7 percent, logistic regression outperforms all other methods on validation data.
Predicting loan defaulters is an important banking practise since it has a direct impact on profitability. However, the loan default data sets provided are extremely skewed, resulting in poor algorithm performance.
This allows us to identify common characteristics and patterns of behaviour that lead to future debt repayment difficulty. For each client, our solution will determine the likelihood of future default.