Comparison of filter approaches based on RVFL classifier

Abstract

Hybrid classification model is currently an active research area and successfully solves classification problems in credit scoring. Finding effective classificatory models is important. Classification in credit scoring has been regarded as a critical topic, with its related departments collecting huge amounts of data to avoid making the wrong decision. Filter feature selection model is important in credit scoring and in the field of data mining. This study proposes three filter approaches which combine with Random Vector Functional-Link net (RVFL) classifier, to find the suitable classification models. Filter approach retains sufficient information for classification purposes. Different credit scoring combinations are constructed by selecting features with three approaches. Two credit data sets from University of California, Irvine (UCI) are chosen to evaluate the accuracy of various filter selection models. RVFL classifiers combine with Grey relation analysis (GRA), conventional statistical linear discriminate analysis (LDA), and F-score approaches as preprocessing step to optimize features space. In this research, the procedures are described and then evaluated by their performances. The results are compared by nonparametric Wilcoxon signed rank test and performed to show if there is any significant difference between these filters. Our results suggest that the performances of the F-score approach combined with RVFL classifier are brilliant among the two data sets. The hybrid model is more effective and higher accuracy than the original feature space and is a promising method in the field of data mining.