Average Reviews:
(More customer reviews)I reviewed the third edition of this book for the American Statistician in 1994. The book covers most of the important topics for an applied course and has a reasonable list of references. There are many examples and homework exercises. Statistical software packages such as SAS and MINITAB are used throughout in example problems. The early chapters cover the basics of statistical inference and regression (Chapters 2-5). This material can be skipped in a first time series course if introductory statistics is a prerequisite.
The latter chapters cover time series regression, seasonal decomposition methods, exponential smoothing and Box-Jenkins methods. But this book does not include nonlinear time series models and it overlooks the recent and popular state space approach to time series modeling. Multivariate time series methods are also left out, though perhaps they are more appropriate for an advanced or second course in time series analysis.
The cookbook nature of the text can be found in the guidelines given for Box-Jenkins model identification. The statistical theory that the methods rely on is avoided. Although a number of important probability distributions are used with their relevant statistical tables, the underlying assumptions and distributional theory is completely avoided.
Important concepts such as the central limit theorem and the concept of a stationary stochastic process are given only very brief treatment. Other concepts are oversimplified to avoid the need for the development of any distribution theory.
This book will serve well for a course in which the student is interested in how to implement exponential smoothing and the general class of Box-Jenkins models through the use of standard statistical packages. However if the instructor wants depth of understanding the text is not adequate. Frequecy domain methods often useful in engineering applications are not even discussed.
While the book covers forecasting applications, it does not consider applications to decomposition of variance or discriminant analysis. Time series methods are also applicable in these contexts. Abraham and Ledolter (1984) "Statistical Methods for Forecasting" cover the same topics but in much greater depth. Also Janacek and Swift (1993) "Time Series: Forecasting, Simulation, Applications" is slightly more advanced and provides broader coverage. Anyone interested in the theory can consult a number of good books including the latest edition of Brockwell and Davis "Time Series: Theory and Methods". Shumway and Stoffer (2000) "Time Series Analysis and Its Applications" is up-to-date, comprehensive and has many good engineering applications.
Click Here to see more reviews about:
Forecasting and Time Series: An Applied Approach (Forecasting & Time)
Click here for more information about Forecasting and Time Series: An Applied Approach (Forecasting & Time)
