Re: [Gretl-users] Using gretl to model ozone datas
El Martes, 7 de Junio de 2005 15:09, Adrien BETON escribió:
Hello
I am a student working on ozone datas and i would like to use an arima
data to correlate ozone datas and gas emissions in order to predict
future ozone levels basing on other gas emissions.
Gretl should allow me to do this but I am not very used to statistics
and modelling.
Therefore I have two questions :
How can I set my datas (setobs) knowing that I have values of ozone each
30' during 6 days (48*6=288 values), gretl only recognize yearly datas ?
The main problem now is that gretl do not support seasonal ARIMAs. I now Allin
is working hard now towards version 1.3.4, and probably this is not in his
inmediate plans, but it is in the "to do" list.
Without seasonal ARIMA, using "Time series/other" to define this type of
series does not have any advantage from defining them as "annual"
What model do you suggest me to use in order to build a model
linking
ozone to gas in order to forecast ozone levels based on the gas ?
The ozone example is one of the classical examples, and you may find
a usefull model in the article of Box and Tiao (1975) in JASA, vol 70, pages
70-79.
If I can t do this with gretl, what other softwares do you suggest me
to
try ?
You may use the "arima" command in R (see http://www.r-project.org ). It is
now in the "stats" package of R. There is a paper by Racine and Hyndman in
the Journal of Applied Econometrics showing some of the econometrics
functions of R. You may find there some examples of the use of the "arima"
command. The title is "Using R to teach Econometrics", and you can find a
working paper version in google.
--
Ignacio Díaz-Emparanza
Dpto. de Economía Aplicada III (Econometría y Estadística)
UPV-EHU
http://www.bl.ehu.es/~etpdihei/