具有高斯过程误差的函数型线性模型的统计诊断(英文)

Statistical diagnostics for functional linear regression models with Gaussian process errors

提出了具有高斯过程误差的函效型回归模型的几种诊断方法.在此模型中,首先,在样条基的基础上,推导了回归系数函数的估计.随后,证明了数据删失模型和均值漂移模型的等价性.然后,研究了三种诊断方法,即残差分析、Cook距离和似然距离来诊断异常和强影响数据.最后。通过一个模拟例子和一个实例来阐述方法的有效性.

Several diagnostic measures are proposed in functional linear regression models with Gaussian process errors.The estimators of unknown regression coefficient functions are first derived based on the spline basis.Then,it is proved that the case-deletion model is equivalent to the mean-shift outlier model.Three diagnostic methods are presented,which are the residual analysis,Cook＇s distance,and the likelihood distance to check outliers and influential cases,and illustrated by a simulated example and a real data example.