I'm not sure this is the right place for this question but i try it anyway.
I cannot reproduce the forecasts in the electricity dataset. According to the article "Forecast Combinations in R using the
ForecastComb Package" in The R Journal Vol. 10/2, December 2018, page 273 they are "1-month forecasts obtained via time series cross-validation".
For example i tried with the cvts function from the ForecastHybrid package to get the forecasts for the arima model, but the forecasts are different than the forecasts from the electricity dataset:
extractForecasts(arimaMod)
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
2007 NA NA NA NA NA NA NA NA NA NA NA NA
2008 NA NA NA NA NA NA NA NA NA NA NA NA
2009 NA NA NA NA NA NA NA NA NA NA NA NA
2010 NA NA NA NA NA NA NA NA NA NA NA NA
2011 NA NA NA NA NA NA NA NA NA NA NA NA
2012 NA NA NA NA NA NA NA NA NA NA NA NA
2013 NA NA NA NA NA NA NA NA NA NA NA NA
2014 32960.93 30855.44 31006.27 27423.08 26918.36 25329.20 25844.95 25965.20 26287.52 28643.59 30341.29 31953.52
2015 33084.71 30680.85 31396.55 27830.51 26561.84 25115.40 25929.45 25682.07 26231.66 28697.33 30345.31 31578.86
2016 32330.54 29612.32 31519.68 27042.02 26844.41 24892.15 25517.59 25138.49 25979.45 28160.74 29292.76 31000.24
2017 32226.42 29701.65 30227.54
There were 39 warnings (use warnings() to see them)
tail(electricity)
arima ets nnet dampedt dotm Actual
Oct 2016 27976.21 28489.29 28282.95 28595.66 28501.43 27117
Nov 2016 28374.38 29940.13 29089.22 30170.96 29853.79 29988
Dec 2016 29526.02 32015.12 31017.64 32360.05 31894.15 30748
Jan 2017 32539.18 32213.54 32392.26 32468.24 32344.26 33248
Feb 2017 31230.25 29746.24 30849.92 29858.74 29685.34 29049
Mar 2017 30466.33 31211.92 30355.76 31324.26 30923.60 28653
I also tried with the tsCV function from the Forecast package but after calculating the forecasts from the forecast errors (which are returned by tsCV) i also get different forecasts then the forecasts in the electricity dataset.
Is the code to recreate the electricity dataset available somewhere or does somebody know how these forecasts were created?
Hi,
I'm not sure this is the right place for this question but i try it anyway.
I cannot reproduce the forecasts in the electricity dataset. According to the article "Forecast Combinations in R using the ForecastComb Package" in The R Journal Vol. 10/2, December 2018, page 273 they are "1-month forecasts obtained via time series cross-validation".
For example i tried with the cvts function from the ForecastHybrid package to get the forecasts for the arima model, but the forecasts are different than the forecasts from the electricity dataset:
I also tried with the tsCV function from the Forecast package but after calculating the forecasts from the forecast errors (which are returned by tsCV) i also get different forecasts then the forecasts in the electricity dataset.
Is the code to recreate the electricity dataset available somewhere or does somebody know how these forecasts were created?