Re: [Gretl-users] using gretl for AR(1) modeling of daily time series
El Viernes, 15 de Julio de 2005 03:22, Allin Cottrell escribió:
On Wed, 13 Jul 2005, Deepak Muricken wrote:
> I am trying to use gretl for modeling daily data as opposed to
> yearly data.
That should be possible.
> I feel that this is possible as one of the sample data (file data
> 10-2), hourly load and temperature data seem to be compatible for
> time series analysis. The data is an hourly information.
That's a bit different: hourly data is "intra-day" rather than
daily.
> What I have with me is daily observation which I wish to use in
> the format given below.
>
> obs (meaning)
> 8:01 1 st of August
> 8:02
> .......
> 8::30 30 th of August
The problem here is that, in gretl, the observation syntax "X:Y" --
where "X" indicates a "major" period and "Y" a
"minor" period --
works only when the frequency or periodicity is constant. For
example, it works if X indexes a year and Y a quarter; or X indexes
a year and Y a month; or X indexes a day and Y an hour.
You may define the data as "daily/7 days a week" and so you will see your data
labeled with dates as you want. The only problem here is that this type of
data has different seasonal depending on which is the periodicity you are
interested on. In most of the cases the relevant periodicity is 7, but if you
have a very long series you may be interested in monthly seasonals, and these
are more difficult to define here as the different months have different
number of days.
In despite of this, I am quite sure you will feel happy defining the data as
"daily/7 days a week".
--
Ignacio Díaz-Emparanza
Dpto. de Economía Aplicada III (Econometría y Estadística)
UPV-EHU
http://www.bl.ehu.es/~etpdihei/