Segmentation Of Retail Customer Based On Data Mining

With the improvement of China’s economic level, retail industry is developing rapidly. However, due to increased competition, the retail industry’s profits reduced day by day, they are urgent to find new profit growth point. For the oversupply of retail industry, how to win and retain customers and maximize customer value has become particularly important, it is also becoming one of the most concern issues of enterprises. C ustomer segmentation is the key to the successful customer retention, the emergence of data mining also provide a new customer segmentation solutions for the enterprises.Existing retail customer segmentation study has two shortcomings: first, there do not form a complete customer segmentation indicator system applies to retails; the second is the problem of low accuracy in customer segmentation model in use. Based on the theory of customer segmentation and data mining techniques, It builds a RFM based multi- indices customer segmentation system, determines the relative importance of the variables by setting index weight using entropy method, and then uses the K-Means algorithm for customer segmentation. Using a supermarket chains’ more than 30,000 customer data and nearly 380,000 sales data to go on empirical research, finally the customers divided into four categories, and giving the corresponding marketing strategies for different customer groups. The empirical results show that this method is better than traditional RFM analysis method on customer behavior characterizing and cluster compactness, and it is feasible and effective and able to better solve the problem of the retail customer segmentation meanwhile improve the quality of customer relations hip management and marketing decisions.Therefore, this study enriches the customer segmentation theory, providing a valuable reference for the future research on customer segmentation. In practice, through the empirical analysis, we have proved the effectiveness of the method, and we can enhance the effect of the model by using the index system for the retail and marketing activities.