| One of the most important departments of insurer, underwriting dept., raining the profit and loss for future to the insurer, has been lift to the position where the dept. never ever arrived.Even if the insurer has kept the symmetrical information with insured as well as possible by the following methods such as agent underwriting, health checking, underwriter and survival investigation by which the more reasonable conclusion can be given, the risk to make loss for insurer is still existed. For example, in present, the risk of suffering disease by insured for future can only be forecasted by experience of underwriter depending on the existed information from insured. It has, however, resulted to some problems like mistaken estimation, excessive relying on experience and not average level of underwriting for each underwriter. Therefore, how to help all of the underwriters to make the estimation more exacted?The thesis submitted Life Underwriting Premium System Based on the Algorithm of RBF-Boosting after researching the problems listed in second paragraph. What is the function is to forecast the risk rate of whether the insured will get disease based on the real insured data found in life system. Firstly, the training records round at 17'000 counts got from life system will be trained by the algorithm of RBF-Boosting. Secondly, another part of testing records 1'000 got from same place with training records will be used to check the veracity of the training results. Lastly, the algorithm will be implemented by Java to join with the life or underwriting system for every underwriter who can get the more accurate forecasting results based on the large columns of the real insured data. Then it could push to make the more adapted underwriting decision for each policy and reduce the underwriting risk further.The core content of the thesis is analyzed further from the underwriting risk requirement of Swiss Life system. |
PDF Full Text Request