Time-series Forecasting by Chris Chatfield
An important general book on time-series forecasting
Published by Chapman and Hall/CRC Press in November 2000.
ISBN is 1-58488-063-5.
This book provides an up-to-date
comprehensive review of forecasting methods. It begins with a
summary of basic time-series analysis and modelling procedures, followed by
a catalogue of the many different time-series
forecasting methods. These range from exponential smoothing,
through ARIMA and state-space modelling to multivariate methods,
such as those based on vector autoregressive (VAR) models.
Recent arrivals, such as GARCH models, neural networks, and
cointegrated models are also covered.
The more important methods are compared in terms of their
theoretical inter-relationships and their practical merits including empirical
accuracy.
The book also considers two other general forecasting topics that have
been somewhat neglected in the literature,
namely the computation of prediction intervals and
the effect of model uncertainty on forecast accuracy.
Although the search for a 'best' method continues, it is now well established
that no single method will outperform all others in all situations -
the context is crucial.
The book provides an outstanding reference source for
the more generally applicable methods, and should be particularly useful to
practitioners and researchers using forecasting methods in
economics, government, operational research, management science, and
commerce. The data sets used in the book are available below.
A few minor typos have been noticed in the first printing.
They have been corrected in the first reprint (September 2001).
Page 18, line 12, deterministic is misspelt. Page 41, l 32, delete 'ensure'.
Page 111, l 17, change 'the' to 'to'.
Page 155, MAPE stands for 'mean absolute percentage error', and
not 'mean absolute prediction error'. This will also be corrected in the index.
Some further misprints have been spotted in both the original printing
and the first reprint.
Page 52, line 4. The G_t matrix is incorrect. First row should be
phi_1, phi_2, while second row should be 1,0 (not 0,1).
Page 94: the superscripts on the alphas on the first line immediately
below the SETAR model given in the middle of the page, should be
subscripts.
Page 149: When trying to write (5.7.1) as a VAR model, the 1st equation
says X_t=X_{t-d}, and NOT X_t=X_t as it should. This model
can only be written as a bivariate VAR(d) model if
we know the univariate model for X_t. This section will have to be rewritten.
If for example, X_t is known to be AR(1), then the VAR(d) model
has a first-order term as well as the d-th order term.
Page 165: FARIMA at end of 2nd to last paragraph should be ARFIMA.
References: Hibon and Makridakis (1999) has been replaced by
Ord, Hibon and Makridakis (2000, J Forecasting, p 443, The M3 Competition)
in both the text and refs. Gomez and Maravall is 2001, not 2000.
Armstrong (2001), Armstrong and Collopy (2001), and
Chatfield, Koehler et al (2001) have all now appeared in print.
The book may be ordered from CRC Press as follows:
In America, Asia and Australasia, send an email to orders@crcpress.com
or telephone 1-800-272-7737 in USA or 1-561-994-0555 from outside USA.
In Europe, Middle East and Africa, you can make enquiries by sending
an email to enquiries@crcpress.com
or by telephoning (44 outside UK) or (0 inside UK) 20 8875 4370.
You can email your order to crcpress@itps.co.uk or telephone the
orders department at (44 or 0) 1264 342932.
Or you can order online at www.crcpress.com
hr>