A Novel Feature Selection Approach based on Multiple Filters and New Separable Degree Index for Credit Scoring
With the rapid development of Internet finance,credit scoring has played a significant role in peer to peer lending platforms.However,the massive and high-dimensional characteristics of the credit data make it difficult to directly build the credit scoring model.Therefore,the feature selection is attracting more and more attention,which can be used for processing the complicated credit data.In this paper,we proposed a novel feature selection approach,which is specifically designed for analyzing the customer data in credit scoring.Firstly,we proposed a strategy that combining multiple filters to select the different candidate feature subsets from customer data.Then,a New Separable Degree index is proposed,which can select the optimal feature subset for credit scoring.Experimental results indicated that the performance of the approach we proposed is superior to other single filters,and the computational cost is greatly reduced compared to the traditional wrappers.
Credit scoring Feature selection Filters Optimal feature subsets
Hongwei Feng Shuang Li Dianyuan He Jun Feng
College of Information Science and Technology Northwest University Xi'an,China
国际会议
2019国图灵大会(ACM Turing Celebration conference-China 2019 )
成都
英文
207-211
2019-05-17（万方平台首次上网日期，不代表论文的发表时间）