Hi Eviews team,
I performed a ETS monthahead daily demand forecast using daily values from May 2016 in order to predict daily values of June 2016.
The output looks very good when I plot the series on top of the actual values. Also the MAPE I calculated manually shows 2.59%, which is very good. But I don't understand why the RMSE on the output table (below) is over 1:
Sample: 5/01/2016 5/31/2016
Included observations: 31
Model: A,N,A
Model selection: Akaike Information Criterion
Convergence achieved on boundaries.
Parameters
Alpha: 0.000000
Gamma: 0.000000
Compact Loglikelihood 264.9415
Loglikelihood 255.7018
Akaike Information Criterion 561.8831
Schwarz Criterion 584.8269
HannanQuinn Criterion 569.3622
Sum of Squared Residuals 26509188
Root Mean Squared Error 924.7352
Average Mean Squared Error 903048.3
Any ideas why?? Thanks so much in advance!
ETS Forecasting high Root Mean Squared Error
Moderators: EViews Gareth, EViews Moderator

 Nonnormality and collinearity are NOT problems!
 Posts: 3289
 Joined: Wed Sep 17, 2008 2:25 pm
Re: ETS Forecasting high Root Mean Squared Error
What is the typical size of your variable being forecast?
Re: ETS Forecasting high Root Mean Squared Error
Hi Chris,
The mean of my series is 28231. Is not supposed RMSE to fall between 0 and 1? Or at least that's what I found among other examples of ETS.
Thanks,
The mean of my series is 28231. Is not supposed RMSE to fall between 0 and 1? Or at least that's what I found among other examples of ETS.
Thanks,

 Nonnormality and collinearity are NOT problems!
 Posts: 3289
 Joined: Wed Sep 17, 2008 2:25 pm
Re: ETS Forecasting high Root Mean Squared Error
Nope. RMSE has the same units as the original variable.
Re: ETS Forecasting high Root Mean Squared Error
Ok, thanks Chris!
One more question, in order to visualize the quality of my forecasting I am calculating scalars involving the forecasted and actual series such as MAPE(%) as @mape(actual,forecasted).
Any other one appart from @rmse that you would recommend?
Thanks again for the help.
One more question, in order to visualize the quality of my forecasting I am calculating scalars involving the forecasted and actual series such as MAPE(%) as @mape(actual,forecasted).
Any other one appart from @rmse that you would recommend?
Thanks again for the help.
Who is online
Users browsing this forum: No registered users and 2 guests