Forecasting: principles and practice

by Rob J. Hyndman, George Athanasopoulos

Forecasting is required in many situations. Deciding whether to build another power generation plant in the next five years requires forecasts of future demand. Scheduling staff in a call centre next week requires forecasts of call volumes. Stocking an inventory requires forecasts of stock requirements. Telecommunication routing requires traffic forecasts a few minutes ahead. Whatever the circumstances or time horizons involved, forecasting is an important aid in effective and efficient planning. This textbook provides a comprehensive introduction to forecasting methods and presents enough information about each method for readers to use them sensibly. Examples use R with many data sets taken from the authors' own consulting experience.

Genres: Nonfiction, Mathematics, Finance, Business, Economics, Artificial Intelligence, Technology

382 pages, Paperback

First published October 17, 2013

About the author

Professor of Statistics, Monash University, Australia

Reviews

[Joshua Hruzik · Rating 4 out of 5]

Forecasting: Principles and Practice by Hyndman and Athanasopoulos is a great intro for time series analysis.

The book covers a wide variety of topics (including dynamic regression and ARIMA) and focuses on the core principles behind these techniques. It is able to avoid the pitfall of being a sole formula collection while it delivers the most important model equations. The language is comprehensible but formal. The great thing about this textbook is its user orientation. Every procedure is accompanied by a full R-script so you can immediately use it for your own research and tackle more complex models.

Some topics would have deserved a more detailed explanation (e.g. exponential smoothing and dynamic regression). If you are completely new to time series analysis you might need some additional online resources to understand these topics.

Oh and don't forget: It's free!

[Bernhard Geiger · Rating 3 out of 5]

This is exactly the type of book that I would recommend to junior data scientists starting with time series analysis. The book introduces a series of quantitative and qualitative methods for time series forecasting, both with intuitive explanations and numerical examples (carried out in R and accompanied by the corresponding code snippets). To keep the material accessible, most of the mathematical details are omitted.

I still decided to reward the book with only three stars for two reasons: First, although I appreciate its simplicity, I had the feeling that some content has been oversimplified to the extent of simply being wrong. Just to give one example, the authors write that "for a stationary time series, the [autocorrelation function] will drop to zero relatively quickly, while the [autocorrelation function] of non-stationary data decreases slowly". This is wrong, as there are non-stationary time series for which the autocorrelation function drops quickly (e.g., a sequence of independent, but not identically distributed random variables), as well as stationary time series for which the autocorrelation function drops very slowly (e.g., a first-order autoregressive process with a defining pole close to one). The second reason for giving only three stars are the last three chapters: Forecasting hierarchical or grouped time series, advanced forecasting methods, and practical forecasting issues. These three chapters consider a lot of different topics (bootstrapping, ensemble methods, backcasting, neural networks, etc.), but each topic is covered too briefly to be able to understand the main concepts behind them.

[Duncan McKinnon · Rating 4 out of 5]

This is a great quick introduction to modern forecasting. The chapters are short and instructive. The drawback is in how high-level the explanations of forecasting models is. It offers a good conceptual explanation of many topics, but lacks the derivations of results and even comprehensive coverage of how modeling methods like autocorrelation and moving averages work. Great book to supplement with a more technical forecasting text.

[dzươn · Rating 4 out of 5]

strangely entertaining. very good material available for free with concrete examples and data sets.

[Stephen Lung · Rating 5 out of 5]

This is a fantastic book introducing time series forecasting covering a wide range of topics (including dynamic regression and ARIMA). Time series forecasting has stumped me for a long time especially trying to integrate machine learning elements into it. However this book has opened the doors into the possibilities to forecast with only a couple lines of code using the tsibble package. As forecasting and more importantly, accurate forecasting is crucial for businesses to succeed in this competitive environment, time series forecasting beyond what's done in Excel is critical.

Recommendation thanks: An individual on my slack channel who went to the 2020 Rstudio conference recommended this book as the best workshop in the whole conference.

Suggestions: Yes but only if you plan on starting to complete time series forecasting by coding. I plan on re-reading this book in the future.

Resource: https://otexts.com/fpp3/

[Pawin · Rating 4 out of 5]

Great book about forecasting with relevant examples. The book provides underlying concepts of forecasting, such as moving-avg, exponential, ARISMA, and etc. Also this book touches the advanced concepts of forecasting, such as Neutral Network, and Autocorrelated Regression. Readers with prior knowledge about Statistics and Math are suitable for reading this book.

[ریچارد · Rating 3 out of 5]

I have not read that many econometrics books and this is just what I was expecting to get and not necessarily what is common with the Authors. There is not that much math which I had no problem with but the examples in the and their discussion book are not building the insight that I was aiming to gain and I don't think the book succeed in that regard.

[Bing Wang · Rating 5 out of 5]

Read the 2nd edition.
Very practical materials. Covered almost everything except state space models.

[Frank · Rating 5 out of 5]

Always good to refresh on this book every few years.

[Corey Runkel · Rating 4 out of 5]

Easy to read and implement, this well-figured book is a good intro to time series and forecasting.

[Bubu · Rating 4 out of 5]

ჩემთვის მარტივად გასაგებად და დაღეჭილად წერია,
იმაზე უხარისხოდ გავარჩიე ვიდრე ვისურვებდი

[Pavel · Rating 5 out of 5]

Hard for me to understand later chapters. Gotta get back someday and reread carefully when needed.

[Satyaki · Rating 5 out of 5]

https://otexts.com/fpp3/

[Daniel · Rating 5 out of 5]

Eminently practical overview of statistical forecasting methods with accompanying R code. Includes a survey of qualitative techniques as well as sufficient technical documentation of quantitative methods without getting bogged down in proofs & derivations.

Excellent resource & reference.

To complete the irresistible value proposition, it's available as a free online ebook via otexts: https://www.otexts.org/fpp

[Adrian · Rating 4 out of 5]

A free online textbook with running code (in R). Very easy to read. Time series made accessible. This what textbooks should be like.

[Leah · Rating 5 out of 5]

This was a great reference, but after working with it for a while I realized the newer version has moved into the tidyverse! Do yourself a favor and go there first

[Maria-Anna · Rating 5 out of 5]

The book is great in covering many themes in the time-series analysis (not just the basic ones) and to provide a mathematical basis for all the mentioned solutions. There were moments when most of the thoughts were provided in code so I probably needed more depth in explaining it in mathematical language but it didn't spoil the impression of the book. It's not an easy read and requires a lot of focus to understand thoughts set out in the book, missing one subchapter may cause an overall misunderstanding of the content of the next chapters as most of the chapters are interconnected.
The only disadvantage of this book that all the provided examples are in R, not in Python but this is just my personal complaint.

[Mel Hanna · Rating 4 out of 5]

Good introduction to different aspects of forecasting with concrete examples in R. There's just enough material here to teach you the right vocabulary and help you decide if you need to deep dive into the topic elsewhere. I do wish there was more explicit discussion of the pros and cons of each approach or when one might be better than another but Chapter 12 on practical considerations was really helpful.

[Rajesh · Rating 5 out of 5]

Fantastic reference for anyone doing time series analysis. Perhaps the de-facto standard among free and public domain textbooks for this area of study/work.

[Nate · Rating 4 out of 5]

Excellent applied guide to forecasting. One of the most practical textbooks I've ever read. Combining the explanations with the code in one book is extremely helpful.

[Ankit Tyagi · Rating 5 out of 5]

A very good introduction to time series with many r codes. Some advance topics are also included but are not that useful. But for the classical forecasting methods this is a good start.

[Pronam Chatterjee · Rating 5 out of 5]

The most comprehensive and lucid book on the art of forecasting with examples in R. Awesome treatment to a fairly complex subject.

[Michael Boerman · Rating 5 out of 5]

hyndman is a LiTeRaL lEgEnD <333333333

[Shelly (YI-Hsuan) LIN]

Excellent introduction of how to implement time series analysis

[Saruul · Rating 5 out of 5]

Great book to get some intuitive concise explanations on forecasting. It comes with R codes.

[Kirill · Rating 5 out of 5]

Many approaches. Misses more math and basics, but helps familiarize with the topic fast.

[Krishna Sangeeth · Rating 5 out of 5]

The definitive book for anyone trying to start learning about Forecasting. Found the book to be a good mix of theory and practical ideas. Read the online version of the book.