In order to investigate feasibility and effectiveness of negative selection algorithm (NSA) in credit scoring, a new model named LR & RONSA Based model, which hybridizingRset optimized negative selection algorithm (RONSA) with logistic regression, was proposed. Firstly, logistic regression selected the most relevant variables with the target variables into the model. Secondly, RONSA used the selected variables as the genes to detect “no-self”. Finally, with experiments on “German Credit set” and ROC curves, it was found that LR and RONSA Based model has better predict effect than the logistic regression model. It can be used as an ideal tool for credit risk prediction.
