Robust designs to model uncertainty with high estimation and prediction efficiency

2016/04/13 by Lin, Chang-Yun
#Computation (stat.CO) #FOS: Computer and information sciences #Methodology (stat.ME)

paper · doi:10.48550/arxiv.1604.03802

Abstract

Alphabetic optimality criteria, such as the D, A, and I criteria, require specifying a model to select optimal designs. They are not model free and the optimal designs selected by them are not robust to model uncertainty. Recently, many extensions of the D and A criteria have been proposed for selecting robust designs with high estimation efficiency. However, approaches for finding robust designs with high prediction efficiency are rarely studied in the literature. In this paper, we propose the P_α criterion and develop its approximation version for two-level designs, called the P_α criterion. They are useful for selecting robust designs with high estimation, high prediction, or balanced estimation and prediction efficiency for projective submodels. Computational studies show that the P_α criterion is a good approximation of the P_α criterion and can reduce great computation time when we search designs over a wide range of models. The connection between the P_α criterion and the generalized minimum aberration (GMA) criterion is studied. Result shows that P_α plays a great role to link the alphabetic optimality criteria and the aberration-based criteria.

Related

• Comparing robustness properties of optimal designs under standard and compound criteria
• A-optimal Designs under Generalized Linear Models
• A reinforced learning approach to optimal design under model uncertainty
• Q_B-Optimal Two-Level Designs
• Robust split-plot designs for model misspecification