Conditional Markov Network Hybrid Classifiers Using on Client Credit Scoring

The client credit scoring is an effective method of credit risk management. At present, the credit scoring of client mainly relies on the weighted sum of each index on index system. However, it results in ineffective using of historical information and the dependency information between indexes. In this paper, a conditional Markov network hybrid classifier is proposed for client credit scoring and existing problem can be avoided. In the classifier, basic index can be discrete or continuous. Furthermore, the quantitative analysis can be accomplished along with customer credit scoring.