Determining the Predictive Accuracy of Loan Defaulters Using R

Abstract

The banking sector shares a great contribution in maintaining the economy of the country. Default in bank loans shares vital role in risk management of various bank institutions. Bank helps the customer by giving many loans, credit cards, investment, mortgage, and others. Bank loan is also an important part of banking institutions which define as the probability of returning money to the bank by its users. Although, with increase in the participant of banking loan user, the number of defaulters is increasing with the minute and that is leading to huge losses for the banking industry. Machine learning has been used to tackle this issue. This research paper proposes a more accurate way to predict loan defaulters. The model proposed in this paper predicts defaulters using a logistic regression area under Roc curve of 77% which beats the earlier accuracy of 75%. Similarly, in this paper, decision trees, random forest, and the SVM models have been able to achieve better accuracy than the models proposed in the past.