Study On Customer Segmentation And Change Mining Based On SOM For4S Shop

With the rapid development of China’s automobile industry, automobile enterprises are facing increasingly fierce competition. With the dramatic growth in car ownership, auto after-sales service market has become more and more beneficial. As a result, the competition among automobile enterprises have already changed from the car’s sales competition to the service competition. As an important sector of automotive companies to maintain the customer relationship, the after-sales maintenance services department of4S shop should accelerate the construction of the customer relationship management (CRM) system which taking the customer as the center and based on the information technology. As the4S shop need to handle the vast customer group which consist of almost all the owners of its brand cars, how to segment these customers is the basis of effective customer relationship management. With the application of data mining technology in the business environment is more and more widely, after-sales maintenance services department also hope to be able to use the advanced data mining technology to support enterprise customer segmentation and customer relationship management policy making.This study views taking data mining technology to the application in4S shop customer segmentation as a system of knowledge discovery process. Based on the characteristic of the automotive maintenance industry, this study developed customer segmentation methods and customer cluster change mining methods. Firstly, customer segmentation indicators were selected based on the characteristic of automotive maintenance transaction data. Secondly the self-organizing maps neural network was applied to the customer clustering, and last the customer segmentation result was generated by processing the clustering result. Thirdly, based on the segmentation result, the customer change over time was analyzed from both customer cluster and customer individual perspective. Some methods was proposed to track cluster number change, cluster attribute change and customer shifts from one segment to another segment over time.