Using random forests to evaluate predicting factors of loan default

Abstract: 

Banks currently use a scoring model to assess whether or not to accept a loan. This model assigns points for lack of derogatory information and deducts points for factors such as bankruptcies, late payments and poor credit history. Even though this system is widely used in the financial industry, using machine learning techniques often results in a more accurate prediction of loan default. This talk will introduce a random forest model which detects the predicting factors of loan default and, using Westpac Bank's historical personal loan data, predicts whether or not a loan will default.

Miriam is an Applied Mathematics Honours student working with Peter Straka and Josef Dick.