非线性再生散度模型基于统计曲率的诊断统计量

Influence diagnostics of nonlinear reproductive dispersion models in terms of curvatures

对非线性散度模型在 Euclid空间建立几何结构 .在此基础上 ,研究了均值漂移模型的曲率度量 .从而导出相应 Cook距离 ,似然距离等诊断统计量的二阶近似公式 .

A geometric framework is proposed for nonlinear reproductive dispersion models in Euclidean inner product space.By using this geometric framework the curvature measure about the mean shift outlier models is studied.The Cook distance and the likelihood distance are obtained in terms of curvature.$$$$