摘要： 根据遗传算法的鲁棒性的特点以及信用评估的模型，文章提出了一种基于遗传算法的信用评估模型，改进了遗传算法，设计了染色体表示方法，用两类客户识别的正确率和错误率来构建适应度函数以及相应的遗传操作，适应信用评估模型的要求。通过对原始数据的训练，得出较为合理的评估模型系数，实验结果表明该方法效果良好，具有实用价值。

关键词: 信用评估, 适应度函数, 遗传算法

Abstract: Based on the generalization characteristic of genetic algorithm(GA) and the credit evaluation model, this paper presents a GA-based credit scoring model, and then improves the algorithm. In order to meet the demand of the model match, chromosome express method, the fitness function using right and wrong rate of discriminating two class customer and corresponding genetic operators are designed. The results show that the proposed method is effective and of great practical value.

Key words: Credit evaluation, Fitness function, Genetic algorithm