The private credit risk prediction represents an assessment of the relative level of risk associated with the timely payments required by the debt obligation.
Credit scoring system that is most common screening method created for the evaluation of new applications is based on the available statistical information that is related to the behavior of former clients with credit.
Usually, credit related business organizations calculate the scoring function from their experiences, so they lack of the reliability with their scoring system.
Another problem is that there are lots of bad debts that was approved as a good. With this matter, credit manager has no reasonable method to capture this dynamically variable status of customer credit.
This paper suggests two step model to solve this problem. First step, application evaluation is to assess applications by statistical function derived from former client*s profile and credit results. Second step, behavior evaluation is to assess the dynamic pattern of the clients trading attributes periodically which is distinguished between good and bad. This paper compares the performance of classification between two advanced statistical methods and artificial neural network.
