APPLYING THE CLASSIFICATION TREE MODEL IN CREDIT RISK ASSESSMENT

Abstract

The aim of the study was to assess the possibility of using the classification tree model as a tool supporting the bank’s operations by reducing credit risk. The question of the possibility of using advanced statistical models as tools to support the activities of banks was presented. A classification tree model was demonstrated to reduce risk in credit decision-making. On the basis of a database of 1,308 successful credit applications (in 2015-2015), described by a set of 27 characteristics of a potential borrower. The model with 93% accuracy indicated statistical regularities allowing to identify a priori credit applicants customers who will fulfill the contract and customers who are likely to have problems with credit repayment. The first criterion for dividing the set of customers into “good” and “bad” was default in the last three years. Other important risk factors were having a limit in the ROR, having an account with the lending bank and the amount of debits declared by the applicant. The age of the borrower and his occupation were also important. Against the background of the results obtained, the possibilities offered by the classification tree model were discussed, and attention was paid to its intuitiveness and ease of interpretation. The benefits of using such models in managing the risks associated with the credit activities carried out by cooperative banks were analysed.

Keywords

• classification tree
• Data Mining
• cooperative banks
• credit repayment
• credit risk