Application Research On Statistical Methods Of Customer Portrait System

In recent years,with the continuous development of the economy,residents' awareness of consumption has become stronger and stronger.Enterprises and companies have a large number of customers,and the customer data is huge.Therefore,all walks of life are faced with such problems.How to make rational use of customer big data,filter and classify different customer groups,and allocate limited resources to the production operations of enterprises reasonably,so as to maximize corporate profits.This paper takes X bank customer management data as the research object,and the relevant variables generated by the customer's actual transaction data are indicators.From the aspects of customer segmentation and customer churn,the data is preprocessed by the synthetic method of principal component dimension reduction and unbalanced samples,and then Customer segmentation portraits and customer churn images were established using SOM clustering and combined forecasting models based on decision tree,KNN algorithm and SVM algorithm.The analysis results show that,firstly,from the perspective of customer segmentation,the customer is subdivided into three categories: high-value customers,medium-value customers and low-value customers.Each category has remarkable characteristics and is consistent with the previous Pareto Principle.High-value customers have the largest transaction volume and the highest customer activity,followed by medium-value customers,and low-value customers have the lowest transaction volume and activity.Secondly,from the perspective of customer churn,according to the model's loss prediction results,Combined with the actual loss situation,the customer is divided into loyal customers,easy to lose customers and lost customers,and from the average account opening time of the demand deposit,this variable can analyze the characteristics of customer churn.Thirdly,after constructing the customer segmentation portrait and customer churn image,the two portraits can be combined to divide the customer into nine categories: high-value loyal customers,high-value easy-loss customers,and then based on customer value.The order and retention order divides nine categories of customers into three gradients.According to the customer portrait constructed above,by analyzing the variable characteristics of each type of portrait customer,a reasonable suggestion arrangement can be given for each type of customer.For the customer segmentation portrait,the value of the customer is mainly from the customer to identify the customer,and Corresponding financial recommendation and marketing suggestions,for the customer loss portrait,mainly from the customer's activity and retention value to classify the customer,and give a reasonable retention strategy for each type of customer.