Mariola Chrzanowska, Master, of, Econ, Dept. of Econometrics & Statistics,, Warsaw University of Life Sciences, ,, ul. Nowoursynowska 166, Warszawa,, 02-787, Poland
The aim of this paper is to evaluate the quality of the client classification (the classification of the clients) obtained by means of artificial neural networks: a multi-layer perceptron and a radial basis function. The selection of methods is not random. The subject of the application of SNN to banking has been undertaken by many authors/researchers (e.g. M. D. Odom and R. Sharda [1990], K. Y. Tam, M. Kiang [1990]). The research is conducted employing actual data regarding the individual borrowers that got a mortgage credit in one of the commercial banks that operate in Poland. Each of the clients is described by 11 variables. The grouping variable informs whether the client pays off the credit regularly due to the credit agreement or he is back in loan redemption. Diagnostic variables describe the clients in terms of demographic features and characterize the credits that are to be paid back (i.e. value and currency of the credit, credit rate, etc.)
The evaluation of the quality of the classification will be conducted on the base of classification errors. The effectiveness of the neural networks will be compared with the results of the prior research conducted by means of aggregated classification trees, since the training of artificial neural trees involves multiple presentation of the training set – just like in a case of aggregated models.