Construction of Customer Classification Model Based on Bayesian Network

At present, the researches on customer segmentation model based on Bayesian network are few. This paper makes a research on the classification problems based on Bayesian network. First of all, it used literature search and case study to describe the related knowledge and classification principles on Bayesian network. After that, combining with Adventure Works Cycles company's customer data, we made the use of K2 learning algorithm to search the best network structure and got two more reasonable Bayesian network topologies. Thereafter, we calculated the posterior probability and selected the largest one of Bayesian network. Then, we adopted 10 fold stratified cross-validation method to verify the correctness of classification model, and the results are satisfactory. Finally, this paper finds out Bayesian network classification has greater advantages than other classification methods.