The Research of Customer Classification Based on Extended Bayes Model

TLDR: An simple and intuitive extended Bayes model was constructed, which can get better classification effect and was applied in customer classification prediction and was veried with examples.

Abstract: Naive Bayes classifier, a classification method based on Bayes theory, shows excellent properties in many fields. In practical application, limited to satisfying independence assumption, it is hard to gain better classification. In this paper an simple and intuitive extended Bayes model was constructed, which can get better classification effect. The extended Bayes model replaces primary attribute group with new attribute group (except categorical attribute) and rearrange weights according to the comparison of the expectation of attribute importance, to reinforce the effect of important attributes and weak the effect of subordinate attributes. The extended Bayes model was applied in customer classification prediction and was veried with examples. Firstly, customers were clustered with K-means algorithm, the cluster result as a pretreatment step was used in customer classification. Prediction by weighted Bayes algorithm, it combines the advantages of the two algorithms to improve the accuracy of customer classification. So a customer segmentation model can be built on a basis of customer lifetime value, customer loyalty degree, client capital credit and etc.