Recent year researches shows that data mining techniques can be implemented in broad areas of the economy and, in particular, in the banking sector. One of the most burning issues banks face is the problem of non-repayment of loans by the population that related to credit scoring problem. The main goal of this paper is to show the importance of applying feature selection in data mining modeling of credit scoring. The study shows processes of data pre-processing, feature creation and feature selection that can be applicable in real-life business situations for binary classification problems by using nodes from IBM SPSS Modeler. Results have proved that application of hybrid model of feature selection, which allows to obtain the optimal number of features, conduces in credit scoring accuracy increase. Proposed hybrid model comparing to expert judgmental approach per-forms in harder explanation but shows better accuracy and flexibility of factors selection which is advantage in fast changing market.