Recommendation Study Based On Customer Value Segmentation

With the current market competition are growing, how to accurately assess customer value for customers , subdivide customers and then provide personalized service has become the enterprise crucial.In the database marketing theory, RFM model is an effective way to measure customer value. It mainly through customer past purchasing behavior preference to predict customer future consumer behavior . In different industries or enterprise, the important degree of three index is different . Therefore, in the application process, we need to find its plus weights according to the actual. Different weights can get different customer value calculation results, thus affecting enterprise customer value accurate assessment and selection of enterprise marketing strategy.But one question is, that we get the R, F, M weights ever by expert scoring, using analytic hierarchy (AHP) method to get the results, but getting the results by experts about weight unavoidably will be influed by subjectivity.This article elaborates the theory and method of customer segmentation.Then based on a retail enterprise sales data and RFM model . Unit time cycle data is Divided into training set and verification collection.This paper firstly analyzes the relations between R, F, M three index and future customer profit for the enterprise within a period of time, then trys to get RFM indexes weights with qualitative analysis under the constraint conditions that the sum of the weight is one. To achieve goals in the current weight training focus, that the possibility of most value customer groups in the future focus remains the most value customer groups is the largest. At this time of RFM index weight is best for this enterprise in this time period, then we calculates the multiple different time period for rules and to find a more excellent RFM index weight. After getting weight, we calculate the customer value and has the different value groups customers classified. Then the different customer group was analyzed, and get the most valuable customer group, then in group within the algorithm for mining association rules in this category customers to consume commodities to get the rules.Finally,we supply the service of recommendations to the customer in order to promote crossover sales and improve customer satisfaction, achieve enterprise and customer win-win situation. Finally, the paper carries on the experimental analysis using sales data of an enterprise.