Analysis Of Customer Churn Of G-Logistics Company Based On Big Data Algorithm

When the logistics industry has achieved rapid development,it has also brought huge challenges to logistics enterprises.The logistics enterprises started late,the foundation is weak,and the second is that the management is immature and the market competitiveness is high,resulting in high customer turnover rate.How to reduce the existing customer churn rate and maintain more old customers while developing more new customers is a key issue facing the current logistics companies.Therefore,this article starts from the practical problem of the loss of logistics enterprise customers,constructing a customer churn prediction model for logistics enterprises,based on the actual customer data of G-Logistics Company,using algorithms such as random forest,through the research and analysis of customer behavior data of G-logistics company,excavate the behavior characteristics of customers who have lost customers,provide early warning to customers who may be lost in the future,and propose customer retention strategies to help logistics enterprises improve the current situation of corporate turnover.The customer churn prediction problem can be regarded as the big data classification problem in data mining.Therefore,this paper will use the big data classification algorithm to predict and analyze the customer churn problem.(1)Firstly,the problem of customer loss in logistics enterprises is clarified.This paper systematically analyzes the status quo of logistics enterprises,the research status of customer churn prediction,gives the definition of corporate churn,and clarifies the research problem of logistics customer churn.(2)Secondly,this paper focuses on the problem of positive and negative imbalance of sample data in the logistics enterprise customer data set,adopts multi-cycle data training set,performs abnormal value,missing value and standardization processing on the data set,and combines SMOTE algorithm tointerpolate the process to G-logistics company customers.The data set is balanced and eventually equalizes the positive and negative sample sets.(3)Then efficiently screen the characteristic indicators.In this paper,we use the random forest gini importance to select the features,select the G-logistics company customer loss data set for feature screening,select 12 characteristic indicators,remove the invalid features,select 10 characteristic indicators,and input the subset into In the SVM model,the model accuracy is improved by 2.29%.(4)Finally,the customer churn prediction model construction and model comparison based on the random forest model.This paper combines data excavator learning to construct a G-logistics customer churn prediction model based on random forest,and compares the model results with linear support vector machine,naive Bayesian and logistic regression model.The experimental results show that based on random forest The customer churn prediction classification model is 9% more accurate than the other models.On the one hand,this paper starts from the problems in the customer churn prediction research,applies the big data algorithm to the customer churn prediction problem,and constructs the customer churn prediction model based on the random forest algorithm.On the other hand,the customer churn prediction problem will be proposed.The actual logistics company is the background,analyzing and modeling its customer data,constructing the customer churn prediction model of the actual logistics enterprise,and carrying out the customer retention strategy for different segmentation values.