Last October I gave a 3-day masterclass on “Forecasting with R” in Eindhoven, Netherlands. There is a follow-up event planned for Tuesday 18 April 2017. It is particularly designed for people who attended the 3-day class, but if anyone else wants to attend they would be welcome.
Every two years we award a prize for the best paper published in the International Journal of Forecasting. It is now time to identify the best paper published in the IJF during 2014 and 2015. There is always about 18 months delay after the publication period to allow time for reflection, citations, etc. The prize is US$1000 plus an engraved plaque. I will present the prize at the ISF in Cairns in late June.
Nominations are invited from any reader of the IJF. Each person may nominate up to three papers, but you cannot nominate a paper that you have coauthored yourself. Papers coauthored by one of the six editors (Hyndman, Kapetanios, McCracken, Önkal, Ruiz, or van Dijk) are not eligible for the prize. All nominated papers are to be accompanied by a short statement (up to 200 words) from the nominator, explaining why the paper deserves an award.
You can see all the papers published in the period 2014-2105 on Google Scholar. You can also download a spreadsheet of the relevant papers with citations as counted by Scopus. Scopus does not cover every published journal, so the citation counts are underestimates, but they give some general guide as to which papers have attracted the attention of researchers. Google Scholar includes far more citations including working papers, but there may be some double counting.
Of course, a good paper does not always get noticed, so don’t let the citation count sway you too much in nominating what you consider to be the best IJF paper from this period.
Nominations should be sent by email to me by 30 April 2017.
In what is now a roughly annual event, the forecast package has been updated on CRAN with a new version, this time 8.0.
A few of the more important new features are described below.
A common task when building forecasting models is to check that the residuals satisfy some assumptions (that they are uncorrelated, normally distributed, etc.). The new function
checkresiduals makes this very easy: it produces a time plot, an ACF, a histogram with super-imposed normal curve, and does a Ljung-Box test on the residuals with appropriate number of lags and degrees of freedom.
fit <- auto.arima(WWWusage) checkresiduals(fit)
## ## Ljung-Box test ## ## data: residuals ## Q* = 7.8338, df = 8, p-value = 0.4499 ## ## Model df: 2. Total lags used: 10
This should work for all the modelling functions in the package, as well as some of the time series modelling functions in the
Different types of residuals
Usually, residuals are computed as the difference between observations and the corresponding one-step forecasts. But for some models, residuals are computed differently; for example, a multiplicative ETS model or a model with a Box-Cox transformation. So the
residuals() function now has an additional argument to deal with this situation.
“Innovation residuals”” correspond to the white noise process that drives the evolution of the time series model. “Response residuals” are the difference between the observations and the fitted values (as with GLMs). For homoscedastic models, the innovation residuals and the one-step response residuals are identical. “Regression residuals” are also available for regression models with ARIMA errors, and are equal to the original data minus the effect of the regression variables. If there are no regression variables, the errors will be identical to the original series (possibly adjusted to have zero mean).
library(ggplot2) fit <- ets(woolyrnq) res <- cbind(Residuals = residuals(fit), Response.residuals = residuals(fit, type='response')) autoplot(res, facets=TRUE)
Some new graphs
geom_histogram() function in the
ggplot2 package is nice, but it does not have a good default binwidth. So I added the
gghistogram function which provides a quick histogram with good defaults. You can also overlay a normal density curve or a kernel density estimate.
ggseasonplot function is useful for studying seasonal patterns and how they change over time. It now has a
polar argument to create graphs like this.
I often want to add a time series line to an existing plot. Base graphics has
line() which works well when a time series is passed as an argument. So I added
autolayer which is similar (but more general). It is an S3 method like
autoplot, and adds a layer to an existing
autolayer will eventually form part of the next release of
ggplot2, but for now it is available in the
forecast package. There are methods provided for
WWWusage %>% ets %>% forecast(h=20) -> fc autoplot(WWWusage, series="Data") + autolayer(fc, series="Forecast") + autolayer(fitted(fc), series="Fitted")
CVar functions have been added. These were discussed in a previous post.
baggedETS function has been added, which implements the procedure discussed in Bergmeir et al (2016) for bagging ETS forecasts.
head and tail of time series
I’ve long found it annoying that
tail do not work on multiple time series. So I added some functions to the package so they now work.
Imports and Dependencies
The pipe operator from the
magrittr package is now imported. So you don’t need to load the
magrittr package to use it.
There are now no packages that are loaded with
forecast – everything required is imported. This makes the start up much cleaner (no more annoying messages from all those packages being loaded). Instead, some random tips are occasionally printed when you load the forecast package (much like
There is quite a bit more — see the Changelog for a list.
We are still looking for a few more invited sessions for the International Symposium on Forecasting, to be held in Cairns, Australia, 25-28 June 2017. Continue reading →
We know Australia is a long way to come for many forecasters, so we are making it easy for you to bring your families along to the International Symposium on Forecasting and have a vacation at the same time.
The International Symposium on Forecasting is a little unusual for an academic conference in that it has always had a strong presence of forecasters working in business and industry as well as academic forecasters, mostly at universities. We value the combination and interaction as it helps the academics understand the sorts of problems facing forecasters in practice, and it helps practitioners stay abreast of new methods and developments coming out of forecasting research.
For the next ISF to be held in Cairns, Australia, in June 2017, we have a great line-up of forecast practitioners discussing some of their forecasting challenges (and solutions). These speakers and their topics are listed below. Continue reading →
Professor Tao Hong has generously funded a new prize for the best IJF paper on energy forecasting, to be awarded every two years. The first award will be for papers published in the International Journal of Forecasting during the period 2013-2014. The prize will be US$1000 plus an engraved plaque. The award committee is Rob J Hyndman, Pierre Pinson and James Mitchell.
Nominations are invited from any reader of the IJF. Each person may nominate up to three papers, but you cannot nominate a paper that you have coauthored yourself. Papers coauthored by Tao Hong or one of the award committee are not eligible for the prize. All nominations are to be accompanied by a short statement (up to 200 words) from the nominator, explaining why the paper deserves an award.
You can see the relevant papers published in the period 2013-2014 on Google Scholar. Of course, a good paper does not always get noticed, so don’t let the citation count sway you too much in nominating what you consider to be the best IJF paper from this period.
Nominations should be sent to me by email by 8 February 2017.
An invited session consists of 3 or 4 talks around a specific forecasting theme. You are allowed to be one of the speakers in a session you organize (although it is not necessary). So if you know what you are planning to speak about, all you need to do is find 2 or 3 other speakers who will speak on something related, and invite them to join you. The length of all such invited talks will be about 20 minutes.
Invited sessions will be marked as such on the program and carry a slightly higher status than a contributed session. Unfortunately, we can’t offer any financial support for these invited speakers or session organizers.
If you are interested in organizing an invited session, please contact us with your topic. The deadline for proposals is 28 February 2017. We don’t need to know who will speak at it — you have a few months to find willing participants after you agree to organize a session.
The ISF is a little different from most academic conferences in that about 1/3 of the attendees are practitioners, and 2/3 are academics. Consequently, we are not only interested in traditional academic sessions, but also in talks from company-based forecasters describing the forecasting challenges they face, and hopefully some of the solutions.
See forecasters.org/isf/ for more information about the conference, and the location. Cairns is one of the most beautiful places in Australia, and very close to the Great Barrier Reef. June is also the best time to visit the area, as it is during the dry season with moderate temperatures and lots of sunshine. We are hoping that people attending the conference will choose to have a holiday in the region as well.
A major news outlet interviewed me on predictive analytics. Here were my responses. Continue reading →