Leveraging Machine Learning to Predict Credit Card Customer Segmentation

Abstract

Explained in this paper is how data mining provides a way to work on distributed Machine Learning (ML) systems, which are already often used in data mining operations. This paper examines eight strategies applied in cases of Taiwanese customer default. The eight methods. Eight distinct classifications are evaluated for prediction accuracy: Random Forest, Naïve Bayesian Classifier, K-Nearest Neighbour, Support Vector Machine, Neural Net, Decision Tree, Logistic Regression, and Deep Learning. Utilizing this method raises the possibility of many consumer loans and is one way to evaluate risk management outcomes, such as the exact probability of credit card loan default. Large financial losses for the borrower could result from default due to the method's overall effectiveness and efficiency. 30,000 Taiwanese clients with twenty-five qualities, all of whom have full payment histories, are the subject of this study's payment data analysis. Four approaches (weighting, SMOTE, Imbalance, and Downsampling) were used to balance the data in this study. We shall contrast four approaches and outline eight distinct approaches in this study.