不同电信欠费率下信用评分问题

Credit Scoring Used for Different Arrears Rates

为提高信用评分的公平性和合理性,研究了电信领域不同欠费率下的初始信用评分问题.在一种电信客户初始信用评分模型的基础上,分别采用遗传算法和蚁群算法,对不同欠费率的客户群体进行数据挖掘,通过评价函数得到最优信用权重分配方案,并对实验结果进行了分析和比较.最后,对原信用评分模型进行了改进,解决了原模型在高欠费率情况下算法解不理想问题.实验结果表明,在采用评分模型进行信用评分时,应针对不同的欠费率群体,可选择不同的信用评分算法.此外,在建立信用评分模型时,需要考虑不同欠费率的情况.

The initial credit scoring of telecom customers for different arrears rates is studied to improve the fairness and reasonableness of credit scoring. Using genetic algorithm and ant colony algorithm respectively, the weights of the customer attributes with respect to the credit are obtained to mine important knowledge from the customer data sets with different arrears rates. A credit scoring model added evaluation factor is proposed in order to find the optimum solution of credit weights in the case of high rate of arrears. Experiments indicate that it is necessary to select different scoring algorithm or different scoring model for the group with different arrears rates.