Plug-in Optimization Method for Penalty Parameters in Nonparametric GMANOVA model
This study addresses the challenges of overfitting induced by basis function expansion and penalty parameter selection in nonparametric GMANOVA models. To this end, a plug-in optimization approach is introduced into the generalized ridge regression estimation framework. By extending this specific plug-in method to the nonparametric GMANOVA setting for the first time, the proposed approach enables efficient computation of penalty parameters and simplifies the algorithm, significantly reducing iterative computational costs. Numerical experiments demonstrate that the method effectively mitigates overfitting and enhances estimation stability. Overall, this work provides a novel avenue for nonparametric multivariate analysis that combines theoretical rigor with computational efficiency.