Analysis On Identification And Development Of Insurance Customer On The Basis Of Data Mining

In recent years, the operating cost of Chinese insurance market has beensignificantly increased while expanding rapidly. On one hand, clients could get moredetailed information, owing to the continuous increasing of insurance market entitiesand channels of distribution, which makes purchasing behaviors more rational.Therefore, the acquisition costs of new clients are continuously increasing. On theother hand, owing to deficiency of the concept of "acquisition costs of new clients"and "maintanance costs of cllients", clients maintanance is overlooked whilepurchasing high increasing rate of clients, which results in the high rate of clientslost. these perspectives directly result in the growth of operational costs of theinsurance industry and the declining of increasing rate. as the gradually increasedacquisition costs and difficulty of new clients, more efforts on maintaning the existingclients are needed while developing new markets of insurance enterprises. It is mostimportant to promote the potential value and the profit contribution of existing clients.the emergence of data mining provides a possible solusion for this issue.This thesis make a discussion on relevant theories and practices of theapplications of data minging in clients development and maintainance. The authormakes a introduction of the concept, classifications, major algorithm and processes ofdata mining. Customer development and maintainance theories in CRM are presentedto realize the combination of data mining and CRM. Then, there are three majorassignments on account of the situation of insurance industry. They are Customeridentification, Customer segmentation and cross-selling.According to the features of data mining, the author uses K-means, C5.0andApriori to analysis the purchase information data of the Customers, which iscombined with the problems of Sunshin Insurance Group Corporation Limited. on thebases of data mining, the author proposes Customer identification model, cross-sellingmodel and Insurance portfolio optimization model. in the Customer identificationmodel, the author concludes the features of the clients who will buy the accident insurance or not in order to forecast the rate of potential clients. in the cross-sellingmodel, the author aims at seeking opportunites to provide new insrurances or servicesto the present Customers. the results of the model are analyzed and reasonableexplanations are also made. Insurance portfolio optimization model is mainly used totestify the rationality of present Insurance combinations and gives some optimizingsuggestions. At the end of the thesis, combined with the main conclusion andproblems of the study, the author points out the deficiencies and future researches ofdata mining in insurance industry.