| In 21 century the insurance companies have been making a rapid progress. The summary of insurance premium is much larger than before. The insurance market competition is more and more drastic and products are more and more similar in quality. Customer Relationship Management (CRM) is a transaction processing and decision support system which based on modern information technology. The insurance company obtains the most economic benefits by analyzing requirments of customers and potential customers'modes chances cost and risk using data warehouse and data mining technology.Data mining, which is defined as the process of discovering paterns in data, is a focus of multidiscipline research. CRM is an application domain of data mining. Clustering is to classify things in term of some essential attributes, which means things of one kind come together. It is one of the most important data mining techniques. Clustering has extensive applications in CRM areas. First, this paper describes the basic theories in CRM, data mining and cluster's analysis . And then this paper introduces in detail a CRM system's design and implementation. A few questions are researched specially, including system's network structure, application logistic structure, data washing, data sheet design, multidimensional relationship anlaysis. The result of experiment shows this system is universal and extensive in CRM System of Insurance.Second, this paper analyses the k-means fast cluster's algorithm in cluster's analysis is deeply studied. This thesis improves the k-means algorithm to reduce the influence of noises and isolated point on clustering result.Finally, provide and apply the improved algorithm to modeling course that insurance customer subdivides. It has realized the application of enterprise's insurance customer subdivide, and help insurance operators carrying on difference marketing to different customer's group, improves the enterprise's key competitiveness of insurance. |
PDF Full Text Request