Distributed Customer Classification Model Based on Improved Bayesian Network

In this paper, a distributed customer classification model based on improved Bayesian network was proposed to solve a distributed customer classification problem. First, using mobile agents which could visit distributed data-sets, the multi-attributes tree and the Bayesian network were built. Then, all the distributed data-sets were trained by Bayesian network structure learning and parameter learning. By this way, customer classification could be evaluated. Comparing with the traditional customer classification models, the experiment result showed that the distributed customer classification model could solve the problems of heavy burden, large storage costs and inefficiency during Bayesian network learning. And this model showed higher forecast precision and better practicability.