Improving accuracy of artificial neural networks for credit scoring models using voting algorithm

Performance of artificial neural network (ANN), one of the useful tools used for credit scoring models, is increased by proposed methodology in present study. Whereas reducing the rate of error, in order to obtaining the best possible result, and optimal network of ANN are very important, in this paper, for reducing the errors of the artificial neural networks, voting algorithm will be offered. Using mentioned algorithm, the outputs of ANN, are categorized in three different groups and according to the taken results, these algorithms have the ability to reduce the resulted errors of the best made model of the neural networks in a value of 1.03 percent.