Algorithmic Analysis And Risk Classification Model In Experiential Ratemaking

The State Council launched a comprehensive automobile insurance reform in the national on January 1, 2003. From January 1,2003 on, China Insurance Regulatory Commission will no longer formulate the unified automobile insurance rate. Company can independently revise and adjust rates. After getting the approval of the China Insurance Regulatory Commission, it can be put into use. Branches can be authorized to conduct car insurance rates "fine-tuning". This reform indicated the starting that the companies get the power of making the rate.Automobile insurance business has a very important position in property insurance, but the existing rate management system can no longer meet the requirements of the current development of the market economy, and is not conducive to the stability of the insurance company, is not conducive to the healthy development of the entire car insurance market, is not conducive to the market mechanism into full play the guiding role. Under the strict supervision of the process of implementation rates, only the large companies can make ends meet after deduct operating expenses, income tax, business tax and increase premiums. Data indicates that some companies operating in the car insurance payment rate at about 57% which is in better condition. Therefore, in carrying out the reform rates, market competition, and scientific orientation, fairly pricing, reasonable classification is the only way to the development of the insurance industry.In determining car insurance rates, the Experience rates system occupies an important position. The main idea is almost same: classify the Customers by Priori information. Make a number of relatively homogeneous risk pools, and the group set a priori premiums; Then adjust the rate after the formation of the premium based on the experience of the insurance claim records for its annual renewal premium. Since experience model based on the risk classification, How to choose effective risk variables and classification model will directly affect the accuracy of the method. Based on the study of the NCD, I introduced data mining technology into the risk classification, and made a detailed analysis. Chapter 1 talk about the experience rates system, focused on the NCD. From the Economic perspective, I researched the existence of the experience rate, which is one of mine innovation. Then I analysed the problems of NCD in practice and the reasons for the problems.In Chapter 2, I analysed the traditional customer risk model and the defect of it, referring to the risk variables selection and risk classification model. Before risk classification, we must first consider the risk classification variable choice. The paper analyzes the flaws of steps and single-parameter model for risk classification variable selection.Chapter 3 is the risk classification model based on data mining, which is the focus of this paper, the contents roughly divided into three parts.Data preprocessing is the prelude of data mining work, and it will occupy most of the work. Through data preprocessing, we can effectively reduce the calculation cost and data retrieval, reduce noise interference and improve the quality of mining. This paper illustrated a number of important studies for the Pre-processing algorithm. Building a data warehouse is an important preprocessing step. Because of their importance and complexity, there is a separate analysis on 3.1.As responses against the second chapter of the shortcomings of traditional methods of risk classification variable selection, in the third chapter, introduce a new risk classification method based on data mining--attributes relevance Analysis. This makes insurance companies can target different areas and the risk characteristics of different clients. Chosen on the basis of empirical data in the most effective risk stratification variable, with a lot of flexibility, accuracy, reduce the subjectivity. It's also one of the innovations.As responses against Chapter 2 of the shortcomings of traditional risk classification models, in the third chapter, introduce a new risk classification method based on data mining-- Decision Tree. Due to the restrictions of customer data, I used a small amount of suppositional data for research and received good results.In Chapter 4, I discussed the application and development prospects of the data mining techniques for domestic insurance industry. I also analysed the difficulties encountered in practice and made some recommendations.Data mining technology has widely use in the field of insurance. With the limitations of time, In this paper, I only analysed the decision tree method. About other data mining techniques'using in insurance can be a follow-up study. Considering the commercial confidential and the privacy of the individual, this article used only a small amount of suppositional data. It is also a major deficiency of this article.