Research On The Technique Application Of Customer Churn Prediction In Auto 4S Store Based On TFM Model

The rapid development of the social economy and the continuous improvement of the transportation infrastructure have led to the rapid growth of Chi na's automobile market.As the main force of automobile service industry,the automobile 4S store have a large number of customers.Compared with the rapid increase in the number of cars,the incomplete infrastructure and the imperfect service measures of the 4S store may cause the customers of the 4S store to continue to lose.As the intangible assets of 4S stores,the continuous loss of customers has brought great losses to automobile 4S stores.Therefore,the research on the application of customer churn prediction technology in automobile 4S stores can provide judgment basis for the managers of automobile 4S stores to develop corresponding personalized services for customers with different values.It provides an accurate basis for lost customers to develop a targeted recovery strategies,and is of great significance to improve the management efficiency of automobile 4S stores and promote the development of customer management in the automotive service industry.In view of this problem,this thesis analyzes the consumption characteristics of 4S stores consumers,and improves the traditional RFM customer segmentation model to form a targeted TFM customer segmentation model for automobile 4S stores.On this basis,the basic attributes of customers,after-sales service needs,customer service and other conditions are studied,and the customer churn model is established to predict whether customers have a tendency to lose.The specific work is as follows:Firstly,based on the traditional RFM model,the indicato rs are improved to form three new indicators: T(TimeRatio),F(FrequencyRatio)and M(Monetary)by analyzing the characteristics of consumer behavior of 4S stores customers.The TFM customer segmentation model is constructed by using T,F and M indicators,and customer segmentation is realized by K-means clustering algorithm.The improved TFM model is able to more effectively segment customers according to the characteristics of customer behavior and provide decision-making basis for automobile 4S stores to segment customers.Secondly,the high-value development customers in customer segmentation results are selected as the research object.Based on the customer churn characteristics of the customer group,the input and output variables of the customer churn p rediction model are determined,and the customer churn prediction model is constructed by using the decision tree method.In order to ensure the efficiency of the customer churn prediction model,the correlation between variables is analyzed by the varianc e method and the chi-square test method,and the input variables are reduced.In the process of establishing the customer churn prediction model,the model is optimized by attribute pruning to improve the efficiency and accuracy of the customer churn prediction model.Finally,in order to verify the validity of the model,this thesis randomly extracts some data from the customer consumption data of a automobile 4S store,conducts experimental verification on the improved customer segmentation method and customer churn prediction model,and evaluates and analyzes the experimental results.The evaluation results show that the improved customer segmentation method can more accurately segment the automotive 4S store customers;The decision tree customer churn prediction model has the best customer prediction effect.