Development of credit scoring model using soft computations

A new approach to construction of the credit scoring model is examined in the article. An approach based on soft computing that allows obtaining acceptable solutions for the ill-structured objects management in the conditions of incomplete, inaccurate baseline information is proposed for the model construction. The advantages of the proposed credit scoring model consist in fact that its construction does not require a priori information about errors distribution. The results of approbation of the developed scoring model on simulated data are presented. In comparison with the known credit scoring models, the amount of training samples can be significantly reduced for the proposed mode.

credit, scoring, credit scoring model, scorecards, economic-mathematical model, regression model, statistical model, neural network, binary choice model, soft computing, fuzzy logic