大型公共建筑工程质量保险纯费率厘定研究——基于GLM及GAM的实证比较分析

为准确地厘定大型公共建筑工程质量的保险纯费率，通过文献分析及调研，对影响大型公共建筑工程质量的因素进行了研究，提炼出11项保险费率因子。以46栋竣工于1993—2016年的大型公共建筑为工程数据集，在频率-强度法基础上，基于Poisson、Negative Binomial、Gamma、Inverse Gauss等分布的假设，利用广义线性模型及广义可加模型建立了8个工程质量保险纯费率厘定模型，通过比较选择最优模型，并对模型进行了验证。结果表明，所建模型可用于大型公共建筑工程质量保险纯费率厘定。

Research on the pure premium rate making of quality insurance for large-cale public buildings:Empirical comparative analysis based on GLM and GAM

In order to improve the accuracy of pure premium rate making of quality insurance for large-scale public building on the basis of literature analysis, this paper studies the influencing factors of quality loss of large-scale public building projects, and extracts 11 insurance premium rate factors. Based on the frequency-severity method and the distribution assumptions of Poisson distribution, Negative Binomial distribution, Gamma distribution and Inverse Gauss distribution, 8 quality insurance pure premium rate-making models are established by using generalized linear model and generalized additive model on the basis of 46 large-scale public buildings data sets completed in 1993—2016 in China, then the best model is selected and its rationality is verified. The results show that the model is suitable for the pure premium rate making of quality insurance for large-scale public building projects.