You are here

The Analysis of Time Series: An Introduction with R

Chris Chatfield and Haipeng Xing
CRC Press
Publication Date: 
Number of Pages: 
Chapman & Hall/CRC Texts in Statistical Science
[Reviewed by
David Gurney
, on
This is primarily a textbook and has been used as such for both undergraduate and graduate students.  Chris Chatfield wrote the earlier editions and, with the assistance of Haipeng Xing, produced the current CRC Press edition.  See our review of the sixth edition.  The book has thirteen pages of references and the index extends to a little more than three pages. 
The book “assumes knowledge of basic probability theory and elementary statistical inference.”   Most of the chapters and two of the three appendices have exercises at the end.  I counted more than 80 exercises and some of these had multiple parts.  The section titled “Answers to Exercises” after Appendix C reveals some answers, but mostly provides notes on how to approach many of the exercises.  A suggestion for an additional exercise would be one on determining the coefficients of the Henderson moving average.  These coefficients are introduced on page 21 without much explanation.
Chapters 1 through 3 concentrate on the basic concepts and methods of time series analysis.  Xing gives a nice review of the existing books on time series in section 1.5.  Chapter 4 and later chapters look at the concepts presented in chapters 1 through 3 in much more depth.
A website has been set up for this book which contains data sets used by the author and the R-code used to produce many of the graphs and results which appear in the book.  Most of the data sets are CSV files, but a few of them are text files.  
The R-code is new to this edition of the text.  In total, there are the almost 70 examples of R-code.  The R-code works as claimed for the most part.   The first graph on page 194 looks like an oscillating function was added to a cosine function with a period of about  \( 100 \pi \), but the graph produced by the R-code does not have that underlying cosine function.  On pages 213-14, the R-code leaves out the definition of the xy function, but the definition of xy is given in the book’s website under R Functions.  Similarly, on pages 329-30, the definitions of x.pos and x.label are missing from the R-code, but they appear in the R-code on the book’s website.   And again, on page 339, the function data2.ccor is not defined, but the definition of data2.ccor does appear on the website. 
The text does not devote any space to explaining the basics of R, but in the book website - if you click on R Functions - there is a link to An Introduction to R from the R website.  A good R function to know is install.packages especially when you try to call up a package like tsDyn and your implementation of R does not recognize it.  The R-code on the website does include a line to install tsDyn, but this is not included in the R-code in the text. 
Looking at the book as a whole, I would say these issues are not too problematic.  The writing is clear and understandable, and the text provides a good introduction to the concepts and implementation of time series analysis.  The frequent use of R-code is quite helpful in demonstrating the ideas discussed in the text.


David R. Gurney is an assistant professor of mathematics at Southeastern Louisiana University where he teaches many sections of elementary statistics.