Personalized Demand Analysis And Classification Research Of Retail Supermarket Customers

Customer personalized demand analysis and customer classification are the keys to enterprise customer relationship management and marketing decisions.Grasping customer needs and accurately judging customer types can help companies allocate resources reasonably,improve competitive advantages,and achieve precision marketing.This paper mainly classifies retail supermarket customers through statistical machine learning algorithms,and proposes marketing strategies for each type of customer.At the same time,after classifying customers,it is more convenient for supermarkets to understand customers' needs,upgrade their products and improve service quality.This paper classifies retail supermarket customers in three steps: the first step is the data processing,the construction of classification indicators,and the contruction of muliti-dimensional classification system on the basis of reference RFM model;the second step is the model method,K-means clustering method is used to classify customers;the third step is to use Fisher discriminant analysis to verify the clustering results in the second step and determine the type of new customers.The univariate and multivariate visual analysis is firstly carried out to analyze the basic situation of customers and their preferences for commodities.Then,K-means clustering is used to classify customers into 5 categories,and charts are used to analyze and summarize the classification results.The distance between different types of centers is far,and the clustering effect is better.Finally,we select the first 70% of the data as the training set,Fisher discriminant analysis is used to obtain the discriminant function and verify the clustering results,and the accuracy of verification is 96.90%,so we reconfirm rationality of the clustering results and the method can quickly judge the category of new customers.