Accelerated Sequential Sampling for Generalized Linear and Ar(p) Models

This note examines the role of accelerated sequential sampling in the case of general linear and AR(p) models. In a general linear model, under independent normal errors, both minimum risk point estimation and xed-accuracy conndence set estimation of the regression parameters are included. Then, the point estimation problem is revisited when the errors are independent, but nonnormal. In the end, the minimum risk point estimation of the autoregressive parameters in a AR(p) model is discussed. For the most part, second-order properties of the accelerated sequential methodologies are emphasized. The paper gives a new methodology for a wide variety of problems in generalized linear and AR(p) models, and it also provides the much needed synthesis of the otherwise scattered literature. This should not, however, be construed as a full review, but it is expected to help readers to arrive at one easily.