| Summary: | 碩士 === 輔仁大學 === 應用統計學研究所 === 93 === It is common to use number deciding method to evaluate the risk of life insurance. When there are more than two risk factors existing on the insured, however, the index must be revised depending on the underwriting experience. If the possible related factors could be analyzed from the historical claim data by statistics methods and information technology, it should be able to advance the approval procedures and quality of the policy of the insurance company.
The main goal of this research is to do calculation and steps of data mining by standard procedures of the CRISP-DM, and to transform and clarify data in order to acquire the appropriate model. The first step is using data description to explore all the information. Then, clustering the insured of different occupations by the Cluster K-Means Analytic method. It can assist underwriters to verify the policy, decrease the short-term claim occurrence rate, prevent adverse selection, improve the quality of the policy, and decrease the policy risk of the life insurance company.
|
|---|
