Analysis of Deviance for Hypothesis Testing in Generalized Partially Linear Models
基于局部多项式拟合，为广义部分线性模型开发了非参数偏差分析工具，提出卡方检验判断非参数项是否显著，并用德国央行数据验证。
In this study, we develop nonparametric analysis of deviance tools for\ngeneralized partially linear models based on local polynomial fitting. Assuming\na canonical link, we propose expressions for both local and global analysis of\ndeviance, which admit an additivity property that reduces to analysis of\nvariance decompositions in the Gaussian case. Chi-square tests based on\nintegrated likelihood functions are proposed to formally test whether the\nnonparametric term is significant. Simulation results are shown to illustrate\nthe proposed chi-square tests and to compare them with an existing procedure\nbased on penalized splines. The methodology is applied to German Bundesbank\nFederal Reserve data.\n