Volatility of income and consumption: Loan default prediction

Abstract

The purpose of the project is to identify which regression model performs better among logistic regression, decision tree and boosting in identifying the defaulters. This study was performed on a set of customers who availed auto loan and aims at identifying the factors contributed to the default. We identified that Logistic regression works better among the three. Introduction: Identifying whether a customer will default or not based on his credentials will be of great help to financial institutions because it helps to reduce credit risk and enhances investment portfolio. In the present competitive scenario where a small mistake can costs you a lot, identifying the potential defaulters on the first hand is of utmost importance to any company. So we tried to develop a model which can predict whether a new customer will default or not based on his credentials. We started with the data set of 33000 rows and 415 variables.