A comparison of data mining techniques in evaluating retail credit scoring using R programming

Retail credit scoring has become more efficient in recent years because of the use of data mining techniques that allow marketing officers and top managers to better estimate their customers credibility. In recent years, many complicated models have been developed; however there are few of them which continues to be used because of its efficiency and simplicity. This study investigates k-Nearest Neighbourhood (kNN), support vector machines (SVMs), gradient boosted model (GBM), Naive Bayes classification, Classification and Regression Tree (CART) and Random Forest (RF) as analytical methods for customer credit scoring estimation and evaluation, using real dataset. At the end of the study it is found that Random Forest model with down-sampling (RF_US) has better accuracy rate when compared to other models.