The Analysis of Time-series: An Introduction with R by Chris Chatfield and Haipeng Xing
A best-selling text on time series.
The 7th edition was published in July 2019 with Haipeng Xing as co-author.
Suitable as a text for a course on time-series analysis for
advanced undergraduate or graduate statistics students.
The new edition has updated examples and uses R throughout
Total sales exceed 50,000.
ISBN is 978-1-138-06613-7 (Hardback) and 978-1-4987-9563-0 (Paperback)
398 pages
This book provides a comprehensive introduction to the
theory and practice of time-series analysis.
It covers simple descriptive techniques, probability models for time series,
estimation, forecasting, spectral analysis, bivariate processes,
linear systems, state-space models, non-linear models and multivariate models.
It includes more recent topics such as GARCH models and neural networks.
The new 7th edition of the book includes new examples, and is updated
and clarified throughout with appropriate R-code for doing the examples, so that
the old Chapter 14 has been removed.
There is a new Chapter 12 entitled 'Volatility Models'.
while the chapter on Multivariate Models has been extended.
Errata to 6th edition: A recent (2013) re-analysis of the yield data in
Section 14.3 has shown that the original analysis used a corrupted data
file (where a few observations were wrongly duplicated). Using the correct data,
some minor changes in values are found, although the structure of the
model identified is unchanged.
On page 282, at 9 lines from the bottom of the page, the autocorrelation
coefficients of
the raw data at lags 1 to 5 should be 0.985, 0.97, 0.95, 0.93 and 0.91,
while the coefficient at lag 24 is 0.52. This still
indicates that first differences should be taken. The autocorrelation
coefficient of the first differences at lag 1 (5 lines from bottom) should
be 0.34. This is still significant.
Of the remaining coefficients, the values at lags
3, 8 and 14 are on the border line of significance (around 0.13) but
their lower value and unusual lag values suggest they should be
ignored. So the same model is identified. The MA(1) coefficient in
the equation on page 283 should be 0.43 rather than 0.37. Likewise,
on page 284, 0.37 should be changed to 0.43 in lines 3, 8 and 9.
The only other significant typos known in the second printing of the 6th edition are:
On Page 48, line 3, the line should start with 'Here Phi(B) = (1 - 0.5B)' (not psi(B)). And in line 8
0.005 should be 0.05.
On Page 310, in the last sentence of the answer to Exercise 6.4, the last expression should have
a minus sign in front.
On Page 323, the reference to Meinhold and Singpurwalla (1983)
should refer to volume 37 (not 32) of the American Statistician.
If you have a copy of the *first* printing of the 6th edition, be warned that
some graph files were accidentally corrupted as follows:
Fig 6.1: Some labels on the horizontal axis disappeared.
The numbers should increase from -10 to +10 in steps of 2.
Strangely the minus signs have appeared.
Fig 6.3: In the right-hand graph, (b) is for beta=MINUS 1 (the minus
diappeared.)
Fig 6.4: (b) is for alpha=-0.7
Fig 7.3: the arrowheads below 'M large' and 'M small' disappeared.
Fig 8.1: the worst case. Nearly all the labels disappeared.
The vertical axes in both graphs should be labelled phi sub xy (omega),
and the horizontal axes in both graphs should go from 0 to pi.
In the LH graph, the vertical axis should go from -4pi to zero in steps of pi.
In the RH graph, the vertical axis should go from -pi to +pi.
These graphs have been printed correctly in the second printing.
If you spot any more misprints, please notify the author.
There are very few typos left in recent printings of the 5th edn.
Page 98: In the line after equation (6.16), f*(omega) d(omega) is
the proportion of variance - the d(omega) is missing.
Page 122: Bottom line. The first word 'form' should be moved 5 lines
further up after 'put in the general'
Tha data used in the 7th edition are available as explained in the Preface, while
most of the data used in the 6th edition is available below:
The 6th edition of the book used the packages Minitab and S-Plus to
analyse data. Most users now prefer a free package called R and this is incorporated into the 7th edition.
Some notes on the use of R are available here
Time-Series Analysis using R
The book may be ordered from your local bookseller or online at www.crcpress.com