Customer Segmentation Analysis Based On Clustering Analysis Method: K-means

At present, corporate management model is transforming from centered-production to centered-customer. In this pattern, the most urgent is to segment their customers and find every customer's or customers'characters. According this result, companies write down their marketing tactics. So, companies must collect a large number of customers'materials and try every means to mine the hidden information from the materials to conduct their practice.CRM is an advanced customer management method. Customer segmentation is one of the most important parts in CRM. One company can segment its customers according to different standards from different facets. In the past, companies selected some character from the customers as the standard to classify its customers. But doing so is primary and inaccurate. Through the segmentation based on clustering, the manager can find some information hidden in the data.In this paper, we take the database of Foodmart2000 from SQL SERVER2000 as the studying object. On the base of building SMC model, the studied problems are mainly the customer value clustering based on the customer profit rate, analyzing and mining the characters of every customer cluster, and the developing direction of the customers of different clusters. So, the main content of this paper is followed as:①Summarizing the theory of CRM, customer segmentation, and the theory of data mining. Discussing several kinds of segmentation methods, the principles of customer and retaining. Introducing the methods of K-means and hierarchy. Finally, confirm K-means as the classifying method in this paper.②Take customer value as the criterion of customer classifying. Through the SMC model. Calculate the present customer value, and use current value to deduce potential value. Then use the two indexes to classify the customers through K-means and distinguish the high-potential-value-high-current customer cluster, the high-potential-value-low-current customer cluster customer cluster, the low-potential-value-low-current customer cluster and the low-potential-value-low-current customer cluster. Then we find the promising customers'development direction of the low-potential-value-low-current customer cluster through the second clustering. ③Test the gained result through the testing data cluster. According to the former clustering result, we analyze and summarize the characters of every cluster. At the same time, we suggest to the managers studying the customers moving between the clusters to constitute the corresponding developing tactics. Doing so is helpful to develop new customers and preserve customers.Eventually, summarize the main jobs in this project and prospect the next direction of the subject.