出版社：Canadian Research & Development Center of Sciences and Cultures
摘要：With the increasing scale of individual credit consumption, the individual credit risk assessment has become more and more important. This paper selected 640 samples from the Germany personal credit database as study object. First, this preliminary screened the primitive indexes, and then used sample classification accuracy as fitness function, making combined assessment model based on linear regression and logistic regression through genetic algorithm. The results showed that the combined assessment model based on genetic algorithm had higher accuracy compared to a single model, and combined assessment model based on the least sum of square error had an advantage in the individual credit risk assessment over the others.
其他摘要：With the increasing scale of individual credit consumption, the individual credit risk assessment has become more and more important. This paper selected 640 samples from the Germany personal credit database as study object. First, this preliminary screened the primitive indexes, and then used sample classification accuracy as fitness function, making combined assessment model based on linear regression and logistic regression through genetic algorithm. The results showed that the combined assessment model based on genetic algorithm had higher accuracy compared to a single model, and combined assessment model based on the least sum of square error had an advantage in the individual credit risk assessment over the others.
