非线性随机效应模型基于统计曲率的统计诊断

Statistical Diagnosis of Nonlinear Models with Random Effects in Terms of Statistical Curvature

该文研究了非线性随机效应模型,为非线性张度大的模型提供了理论和方法。首先在欧氏空间上建立了类似Bates和Watts的几何结构,然后在这基础上研究了均值漂移模型的曲率度量,导出了相应的Cook距离。

We discussed the nonlinear models with random effects,presented the theory and methods for the models with large nonlinearity. First, a geometric framework is proposed for this model in Euclidean inner product space. Then, by using this geometric framework the curvature measure about the mean shift models is studied.The Cook distance and the likelihood distance are obtained in terms of curvature.