Comparison of the primitive classifiers with Extreme Learning Machine in credit scoring

With the rapid growth in the credit industry, credit scoring classifiers are being widely used for credit admission evaluation. Effective classifiers have been regarded as a critical topic, with the related departments striving to collect huge amounts of data to avoid making the wrong decision. Finding effective classifier is important because it will help people make an objective decision instead of them having to rely merely on intuitive experience. This study proposes two well-known classifiers, namely, K-Nearest Neighbor (KNN), Support Vector Machine (SVM), which will be used to find the highest accuracy rate classifier without features selection. Two credit data sets from University of California, Irvine (UCI) are chosen to evaluate the accuracy of various classifiers. The results are compared and the nonparametric Wilcoxon signed rank test will be performed to show if there is any significant difference between these classifiers. Performance of the KNN classifier is better in only one data set but not significant, whereas SVM classifier is significant superior to Extreme Learning Machine (ELM) classifier in the German data set. The result of this study suggests that the primitive classifiers did not achieve satisfactory classification results. Combining with effective feature selection approaches in finding optimal subsets is a promising method in the field of credit scoring.