Analysis Of Retail Shopping Basket Based On Customer Loyalty Classification

With the arrival of the“Big data era”,the attention of retail,medical,e-commerce and finance industries on data mining is becoming more and more important.In the retail industry,the customer is the first productivity,the customer loyalty classification is the important content of the customer relationship management,the shopping basket analysis is an important method to excavate the customer's buying habit.It is very important to increase customer retention rate for retail industry by using data mining technology to solve customer classification and mining customer purchase habits.This paper selects the microdata of the purchase records of 555 members of a large supermarket in 9 years,and analyzes the data mining methods such as cluster analysis and association rules,which mainly includes the following work:(1)In the study of customer loyalty,combined with improved RFM model and the classic wei-dong li(2005)put forward the new model of structure,create index system then using the hierarchical analysis of clustering algorithm and K-means algorithm for different observation individual customer loyalty index clustering,it is concluded that the optimal solution,the final is divided into four types of customers.(2)On the basis of classifying customer loyalty,we conduct shopping basket analysis on four categories of customers:gold,loyal,ordinary and unfaithful.It is concluded that the purchasing power of gold customers is very strong for staple food,snacks,sauce soup and household goods.Loyal customers have less purchasing power in household products than gold customers,and they buy more snacks and drinks products.Compared with loyal customers,the average customer has less purchase of dried fruit and candied fruit,which is more purchased as a staple food.Compared with ordinary customers,non-loyal customers often purchase daily consumption goods.Instead,they buy the rmost expensive non-consumables,such as jewelry.(3)After adding the constraint conditions of the relative rule algorithm Apriori to increase the proportion of sales,it also improves the running rate of the frequent item set.On the one hand,through the sales of this constraint conditions to improve the value of the shopping basket,avoid to produce association rules which has no practical help to improve enterprises sales via a large number of low-cost commodity combination basket;On the other hand,when(k-1)frequent item sets produce k item candidate frequent item sets,the Apriori algorithm can be improved.(4)Comparing the results of shopping basket analysis for four categories of customers,and putting forward corresponding suggestions.For the main purchase of gold customers,we can consider the regular introduction of this marketing strategy to drive sales volume.For snack drinks products with high correlation between loyal customers,the cross-selling strategy can be considered for reducing the price of one person to promote the sales of the other.For regular users of staple food,we can consider bundling the strategy to entice them to buy.For non-loyal customers,it is possible to increase the sales or regular activity information for them to increase the sales of non-consumable items.