Although loan, cash card, and credit card are the important resources of high returns for banks, they can produce high risk of default. As long as some customers don't return the money on time, it brings banks serious dangers. It may not only generate many bad debts but also make banks close down! High rewards and low risk are always the main objective for banks. They hope every customer's credit is excellent and would not cause any loss. If banks can earlier find out bad customers and refuse their requests, the probability of risk will be lowed down. The aim of the research is to construct a prediction model of risk to control the risk and reduce the probability of it by handling some important variables of the data and using Discrimination Analysis technique from the data which comes from bank A. Our model will take the collinear of variables and transformation of categorical variables into consideration at same time.