Abstract:

Using a decision tree classification method, this paper analyzes the data of the customer transactions and customer classification information. The main purpose is to establish the classification model by the classified customer data. The first of model constructing process is a series of basic information processing of data mining process to the collection of information and transactions, in order to reduce its data redundancy, improve its training efficiency and accuracy. Secondly, based on the RFM classification model, the thesis selects the frequency, the average amount of each purchase as the  classified assessment of the supplementary indicators from customer transactions, and thus obtains a group of data training set on customer transactions. Then the opensource data mining tools, the  WEKA and J48 (C4.5) algorithms,are used  to train the customers transaction data training set. Finally, based on testing, certification and verification,a customer classification model is built. And the model of customer classification for customer classification systems benefits prototype system analysis and system design.

Key words: customer segmentation;data mining;classification;decision tree;WEKA