Study On Data Mining Application Of The Insurance Industry

With the national policy support, the Chinese insurance industry has experienced rapid development, the insurance density of 0.46 yuan / person from 1980 to 736.74 yuan / person in 2008. Although the development of domestic insurance industry is fast, but the road is very rough, it is incompatible with level of economic development. Reform and restructuring, resulting in the domestic insurance market has undergone a fundamental change in the environment, changed the pattern of monopolizing the market for individual insurance companies, formed several large insurance companies led a number of small insurance companies involved, the new insurance companies continue to adding a new pattern. Combined with the domestic insurance industry's profitability and management capacity is far behind the developed countries and more open of domestic insurance market to increase the pressure on the domestic insurance business and life. Therefore, fast and good development of China's insurance industry to support economic development and building a harmonious society becomes issues focused by government, industry and academia. Through overview and analysis about theory and application of foreign insurance industry data mining technology, combined with the domestic insurance industry and information management level, this paper presents a data mining technique in the application of the domestic insurance industry.Theoretical research and engineering through experimental analysis, this paper gives the data mining technology in the application of the domestic insurance industry, the feasibility and necessity. Research and experimental results reflect the following two aspects:1) Applications of Data Mining in the insurance industry.(1) Customer relationship management: the application of Data Mining technology in customer relationship management is mainly focused in customer service analysis and customer credit evaluation. Customer service Mining mainly analyze the activities of customers'consulations and compliants, find the basic characteristic of such kind of customers, then serve aimly . Customer credit evaluation is mainly through the Data Mining analisis, extract the information of customer credit grade and relevent customer characteristic, according to the characteristic of different credit grade customer, supply the relevent service and monitor. Customer relationship management is a very widely concept, hence this article just positions CRM to be customer service and customer credit grade. (2) identification of target customers. Target identification of three methods: survey questionnaires, statistical analysis, data mining method. By comparing the advantages and disadvantages of three methods, that data mining is suitable in handling massive data sets in high dimensional space and to identify target customers advantage. (3) cross-selling policies. Data mining can extract characteristics and behavioral characteristics of customers through analyzing a variety of insurance products purchased by them, to provide guidance for insurance marketing. (4) customer retention and loss analysis. Using data mining techniques to extract features of losing customers, build customer loss model, and apply models to predict the probability of loss of existing customers in order to improve specific service or adjustment of product structure. (5) policyholders fraud. The insurance fraud is referring to unilateral fraud in this article, that policyholders'misconduct to the insurance company. Compared with statistical analysis, data mining technology in the treatment of small probability event has obvious advantages. Through data mining of massive high-dimensional data processing, to extract the basic characteristics of client fraud, the establishment of early warning models of fraud, real-time monitoring and assessment of fraud risk policyholders.2) The engineering applications of Data Mining in the insurance industry(1) the basic characteristics and functions of data mining, as well as the mainstream data mining algorithms. (2) data mining system architecture, SAS EM4.3 data mining capabilities and the basic principles of selecting data mining tools. (3) data mining applications in the insurance industry, the general processes and data mining standard CRISP-DM. (4) for the theme " purchase predict to insurance policy " of data mining research and analysis of engineering experiments. Include: data preparation process, index screening methods, data mining method selection, model evaluation , model efficiency and the model deployment of the required attention to such matters. described the basic principles of experiment algorithm: decision tree, neural network algorithms and logistic regression analysis. Comparison of statistical methods and data mining. Pointed out the applicability of each method, advantages and disadvantages, and data requirements. (5)The engineering practice based on SAS EM4.3 and MatLab2007. For the engineering level, showed the detailed application process of data mining tools, including the interpretation of the key code. Analysis of how to embed data mining model to other information management systems to enable them to self-study and updated to improve prediction analysis.