The Research Of Fresh Industry Customer Behavior Analysis Based On Data Mining

With the continuous progress and development of society,enterprises were diversified development trend.the competition between enterprises is becoming increasingly fierce.As a kind of intangible assets,customers are widely concerned by the enterprises,which makes the enterprise's marketing model from product as the core to customer-centered.Therefore,the analysis of customers has become the focus of business content.This is not only conducive to long-term development of enterprises,but also enable enterprises to reduce or avoid the operational risk under the unstable market environment.In the process of rapid development of e-commerce,the analysis of customer behavior appears more and more problems.How to keep the existing customers? How to explore potential customers.How to predict the loss of customers? These problems have been the key issues which e-commerce is facing.How to maintain existing customers,to explore potential customers,as well as the prediction of the loss of customers,etc.,has become one of the key issues which e-commerce is facing.Data mining technology is widely used in recent years.it aims at discovering potential and valuable pattern from database composed of the daily accumulation of large amounts of data.Data mining technology is widely used in customer behavior analysis,which provide decision support for enterprises.In this paper,data mining technology is used to analyze customers in the background of fresh electricity supplier.First,the clustering model is established,and the K-means clustering model is used to classify the customers.The optimal number of clusters k is determined before classification.The evaluation index of the number of best clusters is the DB value,and the smallest DB value corresponds to the number of the best cluster k.At the same time,compared with the result of Hierarchical clustering,the result of K-means clustering is better.According to the evaluation method,k=3 is determined,which means the customer is divided into three types.Based on the analysis of the behavior characteristics of the three types of customers,it is found that the three types of customers are potential category,loss type and loyalty type.combined with the clustering variables,can be as a classification model variables.the decision tree algorithm could generate classification rules.According to the classification rules generated by the decision tree,Enterprises can predict new customer categories,make targeted marketing strategies to provide personalized services to improve corporate profits.