广义泊松回归模型的推广及其在医疗保险中应用
发布时间：2018-07-31 05:30
【摘要】：计数数据往往存在过离散(over-dispersed)即方差大于均值特征,若利用传统的泊松回归模型拟合数据往往会导致其参数的标准误差被低估,显著性水平被高估的错误结论。负二项回归模型、广义泊松回归模型通常被用来处理过离散特征数据。本文以两类广义泊松回归模型GP-1和GP-2模型为基础,将其推广为更为一般的GP-P形式,其中P为参数。此时,P=1或P=2,GP-P模型就退化为GP-1和GP-2模型。文中最后利用此类推广的GP-P模型处理了一组医疗保险数据,并与泊松回归模型、负二项回归模型拟合结果进行了比较。结果表明,推广后的GP-P模型的拟合效果更优。
[Abstract]:The over-dispersed of the counting data is always larger than the mean value. If the traditional Poisson regression model is used to fit the data, the standard error of its parameters will be underestimated, and the significance level will be overestimated. Negative binomial regression model and generalized Poisson regression model are usually used to deal with discrete characteristic data. Based on two kinds of generalized Poisson regression models, GP-1 and GP-2, this paper generalizes them to a more general GP-P form, where P is a parameter. At this time, the GP-P model of PX 1 or PJ 2H degenerated into GP-1 and GP-2 models. Finally, a group of medical insurance data is processed by using this extended GP-P model, and the fitting results are compared with Poisson regression model and negative binomial regression model. The results show that the fitting effect of the extended GP-P model is better.
【作者单位】： 首都经济贸易大学金融学院;
【基金】：2015年中国保监会部级课题项目支持 国家社会科学基金项目(12CJY114)支持
【分类号】：O212.1
本文编号：2154574
[Abstract]:The over-dispersed of the counting data is always larger than the mean value. If the traditional Poisson regression model is used to fit the data, the standard error of its parameters will be underestimated, and the significance level will be overestimated. Negative binomial regression model and generalized Poisson regression model are usually used to deal with discrete characteristic data. Based on two kinds of generalized Poisson regression models, GP-1 and GP-2, this paper generalizes them to a more general GP-P form, where P is a parameter. At this time, the GP-P model of PX 1 or PJ 2H degenerated into GP-1 and GP-2 models. Finally, a group of medical insurance data is processed by using this extended GP-P model, and the fitting results are compared with Poisson regression model and negative binomial regression model. The results show that the fitting effect of the extended GP-P model is better.
【作者单位】： 首都经济贸易大学金融学院;
【基金】：2015年中国保监会部级课题项目支持 国家社会科学基金项目(12CJY114)支持
【分类号】：O212.1
【相似文献】
相关期刊论文 前6条
1 陈小龙;;泊松回归模型[J];科技资讯;2012年17期
2 徐昕;郭念国;;双泊松回归模型在汽车保险索赔次数中的应用[J];河南机电高等专科学校学报;2012年01期
3 戴林送;林金官;;广义泊松回归模型的统计诊断[J];统计与决策;2013年21期
4 王芯;吕晓玲;;应用潜在分类泊松回归模型及EM算法研究网络购物使用次数[J];统计与决策;2011年01期
5 陈异;戴琳;寇鹏;;零膨胀泊松回归模型及其在交通事故中的应用[J];计算机技术与发展;2013年10期
6 ;[J];;年期
，本文编号：2154574
本文链接：https://www.wllwen.com/kejilunwen/yysx/2154574.html
