作者: Robert A. Wolfe , Gina R. Petroni , Catherine G. McLaughlin , Laurence F. McMahon
关键词: Econometrics 、 Deviance (statistics) 、 Poisson distribution 、 Mathematics 、 Statistics 、 Poisson regression 、 Residual 、 Analysis of variance 、 Statistical model 、 Regression 、 Outlier
摘要: We consider methods for selecting the joint specification of mean and variance functions in statistical models rates or counts. Based on analyses diagnosis-specific hospital discharge Michigan, we show that a Poisson model with an extra component systematic variation is superior to several other probability regard error structure. Further, deviance residual appears Pearson residual. The proper such crucial many types analyses, as identification outliers regression designed explain variation.