What standard data sets are used for testing, evaluating and comparing forecasting algorithms? For example, if you're reviewing a paper that describes a new forecasting algorithm, what data set would you expect results for?

For example, there's the Iris Flower database for clustering, and the MNIST data set for handwriting recognition.

Would Rob Hyndman's Time Series Data Library (https://datamarket.com/data/list/?q=provider:tsdl) be a good choice?

I know that questions of the form "Where can I find data that fits these criteria" are off-topic here and that a similar question (Data set for forecasting) has been closed. But it seems that questions about the "gold standard data set" for a particular purpose are allowed (e.g., Require gold standard data for face recognition).


1 Answer 1


I know of no time series data as ubiquitous as the Iris or MNIST data sets however you might want to take a look at the M3 competition data found here. A description of the competition is available here. The data is also available in an R package on CRAN

Another forecasting competition held at IJCNN’04 is the CATS benchmark, which is discussed in Time series prediction competition: The CATS benchmark. A description of the data can be found here and the data is available here


There has been a new M4 Competition run, the data as well as a description of the data is available here.

The M4 consists of 100,000 time series of Yearly, Quarterly, Monthly and Other (Weekly, Daily and Hourly) data.

The minimum number of observations is 13 for yearly, 16 for quarterly, 42 for monthly, 80 for weekly, 93 for daily and 700 for hourly series.

The 100,000 time series of the dataset come mainly from the Economic, Finance, Demographics and Industry areas, while also including data from Tourism, Trade, Labor and Wage, Real Estate, Transportation, Natural Resources and the Environment.

A genereal overview of the M(akridakis) competitions is available on wikiepedia

The Makridakis Competitions (also known as the M Competitions or M-Competitions) are a series of open competitions organized by teams led by forecasting researcher Spyros Makridakis and intended to evaluate and compare the accuracy of different forecasting methods.


Spyros Makridakis, Michele Hibon. "The M3-Competition: results, conclusions and implications". Internation Journal of Forecasting. Volume 16, 2000, pp. 451-476

Amaury Lendasse, Erkki Oja, Olli Simula & Michel Verleysen. "Time series prediction competition: The CATS benchmark". Neurocomputing, Volume 70, 2007, pp. 2325-2329.

  • $\begingroup$ What about Box and Jenkins Air Passengers data set? I would argue that that this equivalent to the IRIS data set. It's famous, it's old, it has interesting features (clear trend and multiplicative seasonality). $\endgroup$
    – Skander H.
    Jul 11, 2018 at 17:35
  • $\begingroup$ Happy to accept an edit with details / links for this dataset $\endgroup$
    – mgilbert
    Jul 11, 2018 at 20:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.