Spectral Analysis of Time-Series Data

Rating:
Summary: An excellent manual for those ignoring everything about spec
Review: I think this book is extremely simple. The only knowledge required to understand it is perhaps Ordinary Least Squares. A theoretical explanation of the spectral analysis is not in the book. There is only a brief comment on De moivre's theorem (the one allowing and sustaining the whole spectral analysis) in a footnote. The main purpose of the author is to deliver an empirical methodology for empirical researchers not interested in the theories. The advantage is that, once you see such empirical applications, you understand the underlying idea of spectral Analysis. The elementary examples are very illuminating. The book is well organized and the review of "de-trending" (on this particular subject, I would like to say that the scientific discussion has evolved a lot in the last 20 years and the author's presentation is a little bit old), "harmonic analysis", "periodogrammes" seems pretty coherent. Perhaps the only drawback is that the text is a little repetitive and thus, slightly boring; but this is a minor problem, if you consider that this style will make more solids the understanding of the fundamental concepts. It's a great introduction to spectral analysis. Students having standard mathematical knowledge should begin here and then start reading more technical works, such as Bloomfied's "Fourier Analysis of Time Series" and the chapter of spectral analysis of Hamilton's "Analysis of Time Series".

Rating:
Summary: Novices Should Start Here
Review: If you haven't got a clue about spectral analysis, this is the ideal place to start exploring the frequency domain. Although the examples are from psychological studies, this is not a reason why an economist shouldn't read it!!!

Rating:
Summary: exceptionally clear
Review: No one writes more clearly than Rebecca Warner. If you want to understand spectral analysis, she will help you. If you want to *do* spectral analysis, read this book first. It might be the only one you need.

Rating:
Summary: A review from Technometrics
Review: The following positive review by Eric Ziegel appears in the November 2004 issue of Technometrics:

Recently, while doing my own research for some application that I was pursuing, I discovered that Technometrics has provided no reviews or reports for any books on spectral analysis during my entire 18 year tenure. So I checked on Amzon.com and found that I had missed a few. I requested a couple of the more recent ones. There have been no new ones for the past few years.

This book was hardly the type of spectral analysis book that I had expected. I learned spectral analysis from the text by Jenkins and Watt (1968), whose two authors were both exceptional mathematicians, which was clearly reflected in their book. Studying spectral analysis always meant dealing with a lot of complex mathematics. Leafing through this book, one will see hardly any equations. In fact, this book presents mostly text. However, that certainly does not diminish its value. Professor Warner has simply devoted an entire book to explaining how to do and use spectral analysis.

The book is focused on sociological, physiological and psychological data. The first two chapters are concerned with planning the studies that will provide the data. A chapter on preliminary analysis (plotting, trend analysis, trend removal) is followed by three chapters that present the basic tools in frequency domain time series analysis. The topics are harmonic analysis, periodogram analysis, and spectral analysis. Chapter 7 summarizes the analysis process and deals with the simultaneous consideration of several different time series. Multivariable time series are limited to the bivariate case, which is the topic for the next three chapters. Here lagged relationships, serial dependence and cross-spectral analysis are the primary topics. The book concludes with one practical chapter, "Pitfalls for the Unwary", and a chapter on "Theoretical Issues".

The analysis of the applications uses the TRENDS program in SPSS. The soft-science focus does not detract from the comprehensive explanations of the practical applications of spectral methods. This book is an excellent companion to the typical, more theoretical book (see., e.g., Stoica & Moses, 1997). Curiously, there are no newer comprehensive books on spectral analysis. Those listed in Amazon.com which are newer are all specific to a particular application such as instrumentation, physics, or communication.

Review by: Eric R. Ziegel, Technometrics, November 2004, Vol. 46(4), pp. 497-498.