Credit Repayment Analysis Using Support Vector Machine And Principal Component Analysis

Abstract

Bank and lenders are required to conduct credit analysis to determine the creditworthiness of customers who applying for credit. These organizations apply a number of different methods in order to perform credit analysis with high accuracy, along with various statistical analysis tools. For this purpose, we will use the German Credit data set which is downloaded from UCI
Machine Learning Repository open access based site. There are 1000 customer records in the data set and the credit status of these customers is encoded with the appropriate ones 1 and the credit status of these customers is encoded with the inappropriate ones 0. In the first step of this study, SVM analysis will be performed using 21 dependent variables and 1 independent variable
in the data set. In the second step of this study, 21 dependent variables will be reduced by performing PCA analysis and SVM analysis will be performed with the dependent variables obtained after the PCA analysis. Will compare the performance of these two different analyzes in the outcome phase of the study.