Page 1 of 1
AR(1) forecasting
Posted: Sun Dec 18, 2016 10:04 am
by MattxLM
Hi,
I have a database of carriers, airport and others variables as degree of an airport from 1999Q1 to 2013Q4.
What I do is estimate an AR(1) model for degree of the carrier "AA" on the airport "MIA" from 1999Q1 to 2010Q4
The equation is named eq_ar in my workfile attached.
When I try to dynamically forecast from 2011Q1 to 2013Q4 in order to compare the forecasting output with real values,it says " Unable to compute due to missing datas".
I have seen that some posts deal with this error but I did not really undestand what to do.
Thank you for your help
Re: AR(1) forecasting
Posted: Sun Dec 18, 2016 10:21 am
by startz
Is your version of EViews up-to-date?
Re: AR(1) forecasting
Posted: Sun Dec 18, 2016 10:22 am
by EViews Gareth
What exactly did you do? (Seemed to work fine for me)
Re: AR(1) forecasting
Posted: Sun Dec 18, 2016 10:47 am
by MattxLM
Thanks for you quick replies.
I have selected the equation "eq_ar" which corresponds to an AR(1) for estimation until 2010 Q4 and selected "Forecast" and just changing the date of the sample as following :
2011q1 2013q4 if carrier="AA" and airport="MIA".
And then it says the error message "unable to compute due tomissing datas" but it creates me the variable degreef whatever.
So i checked the graph I wanted, and it is definitely not what I should have (see the attachement ...)
Re: AR(1) forecasting
Posted: Sun Dec 18, 2016 10:51 am
by EViews Gareth
Check you version is up to date.
Re: AR(1) forecasting
Posted: Sun Dec 18, 2016 11:09 am
by MattxLM
I am using eviews 9 and I cannot update my version because it is a license for university ...
When I try to update it says :
" This feature has been disabled by your administrator "
Re: AR(1) forecasting
Posted: Sun Dec 18, 2016 12:10 pm
by MattxLM
I finally managed to update eviews to the latest version.
I got a new error message, you can see it on the image.
I do not get it insofar as my workfile range is until 2013Q4 and I should not be outside my workfile range..
Re: AR(1) forecasting
Posted: Sun Dec 18, 2016 1:52 pm
by EViews Gareth
If you click ok on the error, does the forecast happen?
Re: AR(1) forecasting
Posted: Sun Dec 18, 2016 2:07 pm
by MattxLM
Yes, the forecast happens and I got this
(The forecasted value is a constant which corresponds to the constant of the ar(1) model)
Re: AR(1) forecasting
Posted: Sun Dec 18, 2016 4:19 pm
by EViews Gareth
Seems like it is working then.