ETS exponential Smoothing Forecasting

[ericos]

Hi!

I am a complete newbie in Eviews!
I am using the automatic exponential smoothing model selection AND forecasting in Eviews 8.
My question concerns the forecast output. Whenever I apply the estimation and a certain forecast endpoint I keep getting this message concerning the forecast output:

"Note that further forecast observations truncated"? This is probably a stupid question but what does that mean?
I need the complete list of actual output in order to compare with another series I have in excel. (I am manually applying MAPE)

Also, my series is extremely non-stationary... Whenever I apply "optimize initial values" the estimation and forecast output gives me the A,N,N (simple ES) over the whole sample range (a flat line throughout). This seems odd to me. If i do not optimize, it gives me a more "expected" output. Is there are reason for this?

(Sorry for my ignorance/incompetence in Eviews)

Regards,
Ericos

[EViews Gareth]

Could you provide the workfile?

[ericos]

here you go!

I should say "if I do not optimize the initial values" as an option in Eviews it gives me more "expected values"...

[EViews Gareth]

What series are you using, what estimation sample, and how many forecast periods?

[ericos]

In "series1", which has 164 observations, I use the "proc" --> ETS ES, and in there I choose as an example:

estimation: "12/31/2010 11/08/2013"
Forecast endpoint: "02/14/2014"

12 cycles (not sure about this one, as I am using weekly data points)
Auto, error
Auto, trend
Auto, season

Model selection: AIC
Model optimization Log-likelihood

Output: multiple graphs and tick all boxes.

[EViews Gareth]

I can't seem to replicate that error message.

[ericos]

I have tried it on several computers, and I still get the same message.

Is there a certain way to put in the dates, maybe my formatting is incorrect.
Could you please explain how you did? Very much appreciate screenshots of your output if possible?

[EViews Gareth]

Here:

[ericos]

Ok thanks.

And in the "forecast comparison table", are you getting all values up until 02/14/2014? I just get up until 01/03/2014 for some reason. And under that table it says "Note that further forecast observations truncated"?

[EViews Gareth]

Ah, now I see the confusion. Yes, the comparison table truncates the forecasts if there are too many observations to fit in the table.

That's just that table display though. If you look at the actual forecast output series, you'll notice you have values all the way.

[ericos]

Thanks for the help, much appreciated!