Decision Tree Methods In Data Mining And Customer Classification

Effective collecting ,maintaining and using customer resources is one of the key questions in the direct-sale enterprises. With the coming of opening the direct-sale market in our country after China's entering WTO, the law of direct-sale will be in effect on 1st oct 2004. From then on ,more and more foreign companies and products of direct-sale will enter China's market, which will be intensively impacted. So how to construct complete management system, especially customer service management system is a urgent problem.This paper study on the objective of customer resource management in the Dalian Shengyuan Biotechnology Company, one of the typical direct-sale enterprise, exploiting decision tree to classify the customer resource, and having mined the ideal customer. The core algorithm of decision tree is ID3, which disadvantage of is easy to select those attributes whose values is more, while attributes whose values is more are not always the best. This paper present a new approach on ID3 algorithm--the information gain degree of attribute--to optimize the decision tree, and obtain rather ideal classification model of decision tree.In additional, this paper supercede the information gain with classification correct ratio during the discreting the continuous attributes, two-element splitting, and simplifying the calculation.