Predicting Credit Paying Ability With Machine Learning Algorithms

Abstract

Most people still have difficulty accessing finance because of a lack or even no credit history. This study aims to develop a data model that predicts a customer's ability to pay from various aspects other than credit history. This study uses the CRSIP-DM (Cross Industry Standard Process Model for Data mining) method. The data used in this study is the Home Credit Default Risk dataset collected by documentation techniques. The data were then analyzed using data modeling analysis techniques, namely logistic regressor, decision tree classifier, random forest classifier, and lgbm classifier. This study found that the best model for predicting client payment ability is the lgbm classifier or the Random Forest Classifier.