Improve Accuracy in Prediction of Credit Card Approval Using Novel XGboost Compared with Random Forest

Abstract

The aim of this work is to conclude the credit card approval using XGBoost algorithm and compare it with Random Forest (RF) to improve accuracy. Prediction of credit card approval using XGboost Classifier with sample size of N=10 and logistic regression with sample size of N=10, and dataset size of 48678. The dataset contains 19 attributes that help to determine whether a person gets approval for a credit card or not. The accuracy of the Xgboost Classifier is 87.97% and loss is 12.03%, which appears to be better than Random Forest (RF), which is 82.86% and loss is 17.14 %, with a significant value p = 0.001 (p<0.05, 2-Tailed) in SPSS statistical analysis. The results show that the Novel Xgboost Classifier seems to perform significantly better than Random Forest (RF) for credit card approval prediction in terms of accuracy.