Research On Enterprise Customer Loss Based On Logistic Regression Model And SVM

In today's era,the highly stable development of enterprises has gradually become a focus of widespread concern,exploring how to make enterprises rapidly develop into a problem of timeliness of equipment.Customers are the foundation of enterprise development,so keeping customers away and absorbing new customers is one of the core tasks of enterprise development.Some scholars use the logistic regression model algorithm to establish the customer churn model through data mining,analyze the key factors affecting customer churn,find out the key factors,take corresponding measures,and achieve good results.However,there are many areas for improvement.The selection of data sets is not extensive,the data set lacks scientific preprocessing,and there are many invalid data,lack of improvement of algorithms,modification of algorithm parameters,and data mining.There are many ways to explore new methods.In view of the above deficiencies,this paper has done the following work.(1)Carefully analyzed the theoretical model of customer churn,and theoretically analyzed how to retain customers and prevent them from being lost.(2)For the data set is large,chaotic and ineffective data,through the custom feature selection method to streamline data,improve data quality,feature selection method,factor analysis),using the statistical data of a communication manufacturer,the original is 1000 Data,42 fields,and finally 1000 data,12 fields.(3)Establish a multi-logistic regression customer classification model,analyze the factors affecting the customer type,and promote different packages for different customer groups,reducing the waste of propaganda.Moreover,according to the key factors affecting the customer's choice of the package,the customer can better introduce the appropriate package to attract and retain the customer.(4)Establish a customer-based churn model based on binomial logistic regression to find the key factors affecting customer churn.In the work,we can make more effective and more favorable benefits for these key factors,in order to attract more customers.(5)Establish SVM-based The customer churn model,the logistic regression model is improved on the algorithm,using the new algorithm,it is found that the SVM model is used to establish the customer churn model,and the kernel function selects the Sigmoid function to enure,equip,internet,age,voice,employ,ed Eight factors such as ininc are input,and the predicted customer loss is the most accurate.