【作者】 朱晓谦； 李建平；

【Author】 Zhu Xiaoqian 1,2,Li Jianping2 (1 School of Management,Graduate University of Chinese Academy of Science,Beijing 100190) (2 Institute of Policy and Management,Chinese Academy of Science,Beijing 100190) ?

【机构】 中国科学院研究生院管理学院； 中国科学院科技政策与管理科学院研究所；

【摘要】 原始的TOPSIS排序方法需要人工主观赋权,在被一些学者改进运用到信用评分当中时,预处理式(Preprocessing)的赋权方法如回归,层次分析法等被用来替代人工赋权。预处理式赋权和信用评分是两阶段式的,二者之间没有直接关系。本文提出了一个包裹式(Wrapper)的信用评分方法,该方法基于TOPSIS的基本思想,在方程中引入各个指标的权重,优化方程时可自动求解出指标的权重,因此命名为AW-TOPSIS。最后,本文给出的算例验证了该方法的可行性。更多还原

【Abstract】 The weights in traditional TOPSIS sort algorithm are determined subjectively.When some researchers apply TOPSIS into credit scoring,they try to determine the weights of attributes in an objective way by using regression method and so on.These weighting methods are the preprocessing method and have no direct relationship with the following credit scoring model.In this paper,we propose a credit scoring model which can automatically attain the weights in the optimization process,so the weighting is belong to wrapper method and the model is named AW-TOPSIS.The outputs of the AW-TOPSIS are classification results and the weights of the features.At last,an example is used to demonstrate the feasibility of AW-TOPSIS.更多还原