Research On Automobile Service Company’s Customer Churn Prediction

Nowadays, China’s automobile industry develops rapidly, also led to a rapidexpansion of automobile service industry. Although there is a great developmentpotential of the automobile service industry, but overall this market started late,service measures need to be perfected, workers’ quality are very low in this field.Meanwhile, this industry lacks of management ideas and technologies, service varietyis single. The way of customer choose product and service becomes various and theloyalty of customers are lower. The customer churn problems already become one ofthe most matters of concern in automobile service industry. Therefore this paper hopesto find a way to build a customer churn prediction model, identify the behaviors of thecustomer churn. In this way, do effectively targeted marketing for different customer,to reduce customer churn, enhance the competitiveness of enterprises.The main work of this paper:1) Confirmed the customer churn problems of the automobile service industry.In this paper, through reviewing the customer churn theory literatures, analyzesthe reasons of customer churn, grasp the research process of customer churn, knowthe necessary of the customer churn management. On this foundation, combining withthe characteristics of the automobile service enterprises, gives the definition andclassification of customer churn problem in automobile service enterprises. Soconfirmed the customer churn problems in automobile service enterprises.2) Designed the framework of automobile service enterprise’s customer churnprediction model.This paper firstly discusses the feasibility of using data mining methods toestablish the automobile service enterprise customer churn model. Through studyingthe application of prediction algorithm in other industries, discusses the principle andsignificance of the automobile service enterprise customer churn modeling. It isinnovated to combining Decision tree and Logistic regression model as a method ofmodeling the customer churn prediction problem in automobile service enterprises.3) Constructed the customer churn prediction model of Shanghai Dongpuautomobile Sales and Service CompanyBased on the date of Dongpu Company. In considering and analyzing otherrelated industry customer churn prediction based on analysis by the use of variablefactors, identify the11closely static customer information of customer churn, such as Service_length, Age, Channels, Saving_numbers and so on. At the same time,introduced the time series factors, finally put this part two factors as input variables ofcustomer churn, the final input of explanatory variables is to167. Then use the crosstable analysis and decision tree method to select the variables, finally use theLOGISTIC regression model for data mining in SAS modeling platform.After thereturn operation decision, to determine the effect of automobile service enterprisecustomer churn of the7largest contribution variable to predict the final conclusion,and thus churn prediction model has builded.4) Applied the customer churn model.By using this model, calculated the churn probability for Shanghai Dongpuautomobile sales and service Company. Use the RFM model to make a effectivemarket segmentation of customer. Finally, combined the customers’ retain theory,designed the control and reduce strategy for customer churn.In this paper, it is the first time to design the framework of customer churnprediction model of automobile service industry, innovated to combining Decisiontree and Logistic regression model as a method of modeling the customer churnprediction problem in automobile service enterprises, and make a great effects. At thesame time, combined the results of customer churn model with RFM model forcustomer segmentation, to achieve the target marketing, make it real to reduce thecustomers churn. This paper has opened up a new perspective of customer churnresearch in automobile service industry.