Designing Experiments for Adaptively Fitted Models

摘要

Stepwise regression is a commonly used tool if the order of the linear model to be fitted is not known exactly. In this note strategies are proposed for an optimal choice of both the design and the decision rule which selects the adequate model.