The Study and Application on Multi-dimension and Multi-layer Credit Scoring

Scoring of customer's credit is the basis of making an investment. Hence how to calculate scores, namely 'Credit Scoring', is an important and difficult task. Based on current methods, a credit scoring method composed by multi-layer analysis is proposed in this paper, which includes removing outliers, clustering, k-Nearest Neighbor. Especially, this method first separates outliers from data set, and then clusters by fuzzy and similarity in order to divide data into uncertain data and certain data. Here, we focus on analyzing the uncertain data to improve the accuracy of our credit scoring. At last, experiments are used to validate our method more efficient than other credit scoring methods.