To determine the performance of a particular developed model, it is necessary to test using available data. In this example, testing of the developed model is carried out using the Matlab program. To determine the borrower's credit rating, a list of fictitious data was submitted to the system entries, which allows one to consider the influence of certain parameters on the value of the credit rating.

Table 3.1 presents the financial condition of the borrower (Finance) depending on the three input values as: salary, length of service and job status.

Table 3.1 – Borrower Financial Variable (Finance)

Table 3.2 shows the indicator Demographics of the borrower (Demographics), which depends on the age, education, marital status and number of children of the borrower.

Table 3.2 – Variable Demographic position of the borrower

Table 3.3 evaluates the Credit history, Financial condition (Finance), Demographic position (Demographics) and the Availability of real estate to determine the final value of the Credit rating of the borrower.

Thus, determining the credit rating of the borrower based on fuzzy logic allows more objective to draw intermediate and final conclusions, highlight the strengths and weaknesses of the client, as well as make suggestions on the advisability of issuing a loan, which directly makes it possible to reduce credit risks.

Table 3.3 – Variable borrower credit rating

The fuzzy logic technique is one of the most important machine learning methods used to assess credit risks. The use of fuzzy logic in the financial sector allows us to solve problems that are often influenced by many different and complex factors that cannot be labeled with a clear number or process. The formation of a rule base for a specific task is one of the difficult steps in the application of fuzzy logic. Despite the complexity of creating a rule base, models based on fuzzy logic can quickly and efficiently solve complex problems, which is uncharacteristic of traditional probabilistic mathematical models.