Bayesian forecasting and dynamic models. (English) Zbl 0697.62029

Springer Series in Statistics. New York etc.: Springer-Verlag. xxi, 704 p. DM 118.00 (1989).
This monograph is a comprehensive text intended for advanced undergraduate and postgraduate students in statistics and econometrics. It discusses in both applied and technical details structure and theory of Bayesian learning and forecasting. Its sixteen chapters can be broadly divided into four parts.
Part one deals with simple dynamic regression models and the problems of learning, monitoring and forecasting in simple environments. Part two deals with the dynamic linear models (DLM) with normally distributed errors, where the Bayesian methods of forecasting, feedforward intervention and monitoring are analyzed with numerous illustrations. Part three discusses time series models, both univariate and multivariate, multiple regression DLM and the seasonality aspects. Part four deals with the exponential family dynamic models, multiprocess and other nonlinear models.
The book provides valuable empirical illustrations and useful algebraic derivations of Bayesian forecasts under alternative assumptions of probability distributions. It is a valuable addition to a textbook library.
Reviewer: J.K.Sengupta


62F15 Bayesian inference
62P20 Applications of statistics to economics
62-02 Research exposition (monographs, survey articles) pertaining to statistics
62A01 Foundations and philosophical topics in statistics
62M20 Inference from stochastic processes and prediction