The 37th annual International Symposium on Forecasting will be held in Cairns, Australia, from 25-28 June 2017. We plan to hold some workshops on Sunday 25 June, before the main conference. Continue reading →
I’m currently in the Netherlands for a few weeks, and I’ll be giving a seminar at the Data Science Centre in Eindhoven next Wednesday afternoon on “Visualization of big time series data”. Details follow. Continue reading →
A common problem is to forecast the aggregate of several time periods of data, using a model fitted to the disaggregated data. For example, you may have monthly data but wish to forecast the total for the next year. Or you may have weekly data, and want to forecast the total for the next four weeks.
If the point forecasts are means, then adding them up will give a good estimate of the total. But prediction intervals are more tricky due to the correlations between forecast errors.
I’ve pushed a minor update to the forecast package to CRAN. Some highlights are listed here.
It has been well-known since at least 1969, when Bates and Granger wrote their famous paper on “The Combination of Forecasts”, that combining forecasts often leads to better forecast accuracy.
So it is helpful to have a couple of new R packages which do just that: opera and forecastHybrid.
Regular readers will know that the International Symposium on Forecasting is coming to Australia in June 2017. This is the leading international forecasting conference, and one I’ve attended every year for the past 17 years.
It will be held in Cairns, Australia — one of the most beautiful locations in the country (and there is some stiff competition!) and right next to the Great Barrier Reef. Some further information is available on our website (still in progress).
This is only the second time it has been held in Australia, with the 2004 conference being held in Sydney. We expect to get about 300 people attending, 2/3 from academia and 1/3 from business, industry and government.
Right now, I’m looking for organizations who wish to get involved with some sponsorship. Sponsor information is highly visible at the conference, as well as on the website, the program and other publications, so it is an opportunity to support the forecasting community, promote your organization, and perhaps recruit some young rising stars in the analytics world. Continue reading →
This is only directly relevant to my Monash students and colleagues, but the same idea might be useful for adapting to other institutions.
Some recent changes in the rmarkdown and bookdown packages mean that it is now possible to produce working papers in exactly the same format as we previously used with LaTeX. Continue reading →
I have a new R package available to do temporal hierarchical forecasting, based on my paper with George Athanasopoulos, Nikolaos Kourentzes and Fotios Petropoulos. (Guess the odd guy out there!)
It is called “thief” – an acronym for Temporal HIErarchical Forecasting. The idea is to take a seasonal time series, and compute all possible temporal aggregations that result in an integer number of observations per year. For example, a quarterly time series is aggregated to biannual and annual; while a monthly time series is aggregated to 2-monthly, quarterly, 4-monthly, biannual and annual. Each of the resulting time series are forecast, and then the forecasts are reconciled using the hierarchical reconciliation algorithm described in our paper.
It turns out that this tends to give better forecasts, even though no new information has been added, especially for time series with long seasonal periods. It also allows different types of forecasts for different forecast horizons to be combined in a consistent manner.
A few years ago, I wrote a paper with George Athanasopoulos and others about a tourism forecasting competition. We originally made the data available as an online supplement to the paper, but that has unfortunately since disappeared although the paper itself is still available.
So I am posting the data here in case anyone wants to use it for replicating our results, or for other research purposes. The data are split into monthly, quarterly and yearly data. There are 366 monthly series, 427 quarterly series and 518 yearly series. Each group of series is further split into training data and test data. Further information is provided in the paper.
If you use the data in a publication, please cite the IJF paper as the source, along with a link to this blog post.