Designing Credit Rating Model for Legal Clients of Commercial Banks through Utilizing WARD and with Fuzzy TOPSIS Approach

ABSTRACT :At the present research we study a credit rating model for legal clients of commercial banks through data mining and with fuzzy approach. For this reason Basel II standard and Internal Rating-Based model (an advanced and fundamental model based on internal credit rating) are used. We study all pursuable profiles related to 53 legal clients of a reliable commercial bank through a research incensusmanner from 2007 to 2011. For model designing, through adjusting independent variables of the research with Basel II standard, and with regard to economic, political and social characteristics of the region totally 54 variables are identified. Considering Basel II definition of rating and credit risk, the Expected Loss (EL), which is obtained by multiplication of PD, LGD, EAD and M, in company with independent variables are used in model designing. These variables along with collected data are regulated in Excel and calibrated with Fitch standard; and then they, as an input, are analyzed (data mined)with SPSS19 software and through stepwise and forward regression method. The variables are divided into four categories. From 55 input variables we select 17 variables with 95% credit. Then through Fuzzy descaling as well as calibration of variables with Fitch standard, the selected variables, as an input, are placed in TOPSIS stages by means of Excel. 53 companies are rated and divided into 7 categories, the 7th of which is merely for defaultclients.