Abstract
It is the first pass to guard against credit risk that commercial bank identified the credit risk of the customers. Firstly, it introduced the theories and methods of bank customer credit risk identification basing on the support vector machines (SVM) classification, and used the index entropy weights screening methods to screen the index of bank credit risk. Using the SVM technology, exampled by one commercial bank credit risk identification, selecting 58 items of credit risk data about its real estate customers, is to detect the credit risk situation. On this base it further discussed how to extend the SVM to multi-class classification, in order to directly carry on the customer’s credit rating based on its identification. It discussed how to employ the incremental learning method, on the basis of keeping the former study results, only to study the newly-rising data, which formed a continuous learning process and offer technology supports to settle the problem of the bank customer credit evaluation on real time.