Model Credit Scoring Menggunakan Metode Classification and Regression Trees (CART) pada Data Kartu Kredit

Abstract

Credit scoring is a tool and prediction technique that helps financial institutions to lend. The purpose of credit scoring is to assign prospective customers or customers to one group of "good customer" or "bad customer ".One method that can be used to evaluate credit scoring is Classification and Regression Trees (CART).Classification and Regression Trees (CART) is a statistical method used to perform classification analysis. This paper discusses how to model credit scoring using the Classification and Regression Trees (CART) method.The calculation of credit scoring data is based on the credit history data of the customer.In this paper the data used are credit card customer payment data from April 2005 to September 2005 in Taiwan. The influential independent variables amount of bill statement in April (X₁₇), amount paid in May, 2005 (X₂₂), the repayment status in May, 2005 (X₁₀), the repayment status in July, 2005 (X₂₀), the repayment status in Agusuts, 2005 (X₇) And the repayment status in September, 2005 (X₆).In this method the classification of credit customers by Classification and Regression Trees (CART) method gives 78.4 percent classification accuracy for training data and 78.6 percent for data testing.