Forecasting and time series books

People often ask me for recommendations on forecasting books and time series books. So here is list of eight good books to which I often refer.

(Updated 8 November 2017)

Two are my own books of course (after all, I wrote them because I thought I had something to say). You can also get the first one online for free at OTexts.org/fpp2.

The book by Ord, Fildes and Kourentzes is excellent, although ridiculously expensive, and is more detailed in some areas than my book with George Athanasopoulos. It has particularly good coverage of the non-statistical parts of forecasting including judgemental forecasting and forecasting within organizations.

Diebold has a more econometric focus than the other books and so has more detailed discussion of unit roots, volatility forecasting, etc. It also discusses forecasting loss functions, which all of the other books ignore.

Pena, Tiao and Tsay contains chapters by different authors, and covers several topics that everyone else ignores such as different types of outliers, Bayesian analysis, nonparametric time series analysis, VARMA models, and more. It is at a higher level than the other ones listed here.

Box, Jenkins, Reinsel and Ljung is the classic and original reference. But it is still worth reading for its deep insights. The most recent edition has been updated with R code, but the quality of the code is particularly poor. Read it for its statistical insights, not for its coding quality.

Armstrong’s “Principles of Forecasting” is by a range of different authors and the chapters are of variable quality as a result, but it is an excellent resource, especially on the non-statistical areas of forecasting.

Finally, Shumway and Stoffer is a good a book on time series using R. It is not great on forecasting, but quite good on other aspects of time series analysis.

comments powered by Disqus