The Comparative Study of Data Mining Clustering Algorithms to Measure Customer Value in Customer Relationship Management in the Insurance Industry

In today's competitive world, there is a significant improvement in enterprises’ relation with their customers. In modern business environment, there is a shift from product based frameworks to customer based frameworks. A thorough understanding of customers and their behaviors has found an important role in business strategies. One of the challenges facing organizations is to identify customers and allocate resources to them according to their values for the organization. As a result, finding a proper measurement and ranking method for different categories of customers is considered as one of the main priorities for customer relation systems.The main goal of the current research is to conduct a comparative study among the clustering algorithms and employ them to evaluate customers’ value (the lifetime value), subsequently. Insurance industry and more specifically Comprehensive Vehicle Insurance is the main target of this research. The proposed methodology for the current research is based on CRISP-DM. Besides that, one of the extended RFM models is utilized in order to categorize customers and measure their lifetime value in each sector. At the first step of this research, the information of nearly 812 customers had been extracted from 4800 data records of selected insurance companies in the period of 1391 to 1392 based on the Iranian calendar. Then, customers were clustered using sophisticated software tools in the area of statistical process and data mining including RapidMiner, SPSS and Matlab. After analyzing the results of each cluster, customers were categorized and ranked based on their lifetime value using silhouette and SSE indexes. Finally, the results of best clustering method were compared and analyzed. The proposed method of this research can be employed to find the value of each group of customers in the insurance industry and to determine the most proper marketing strategies for each of them.