I have time-series data with a large number of samples (anywhere from 180 to 360 daily samples). I would like to use GRETL to automatically pick a ARIMA model to forecast values for as many as 30-60 days into future. I would like to also model for seasonality in the data.
I have been playing with gretl tool to explore my data with various ARMIA models - like ARIMA (1,0,1), ARIMA (1,1,1) and it has been a great experience so far.
Now, here is my issue. I would like to write a library to basically take time series data as input, try various ARIMA models and pick the most appropriate one "automatically" without human input or intervention. I understand this is fairly novice way to looking at the forecasting, but my application's use-case is simple enough to do it this way.
Now, has anyone tried this using GRETL where multiple models are tried and results (fstats output) are compared to pick one over the other? If so is there a GRETL script out there that can be shared?
Thanks in advance,
Sorry, this is a bit long but I hope you'll bear with me.
First, I'd like to second a comment made by Jack Lucchetti on the
gretl-devel list, namely that the idea of announcing a pre-release
bug-hunting period has proved its worth very nicely. Thanks to
Ignacio Diaz-Emparanza for proposing this, and thanks to all who
have submitted reports. Hopefully, gretl 1.9.6 will be a "keeper".
Second, in the process of responding to bug reports I've modified a
few translatable strings, so I propose to wait a few more days
before releasing to give translators the opportunity to catch up.
I'm now expecting to release gretl 1.9.6 by the end of this week.
Third, in the spirit of the bug-hunting period I have tried to
restrain myself and not add new functionality that might pack new
bugs. However, there are a few new things in CVS that may be worth
1) Adding an importer for new-style Excel files (Office Open XML,
*.xlsx) was less difficult than I feared, so this will be in 1.9.6.
2) I have a new econometrician colleague who has been learning gretl
fast, and in writing a simulation script he noted some room for
improvement in our random number generation functions. In response I
have a couple of changes:
(a) Added a new function mrandgen() which provides the same
functionality as randgen(), but with matrix output.
(b) Added a new distribution option, 'i' to randgen and mrandgen:
this gives discrete (integer) uniform output over a specified closed
interval. And there's also a new function randint() which does the
same but with scalar output.
It's possible to write your own function to generate random integers
on the basis of gretl's continuous uniform output, but it's not
trivial to get that right; it's also rather roundabout, since the
underlying RNG used by gretl offers integer output natively. The new
discrete random functions give direct access to the integer RNG.
Fourth, I have just recently tried to improve the behavior of the
GUI script editor in respect of the Tab and Enter keys. The idea is
that "smart tab" at the start of a line should produce the correct
indent, and Enter should correct the indent of the line that's being
terminated, if need be. Up till now this has worked up to a point
but has not been fully reliable, and I think it should now be
better. Please let me know if you see any anomalies.
I'm having some troubles with graphs' GUI zoom command :
After a 1st zoom I replace the full view and make a 2nd zoom but the
result seems not consistent.
(axes happen to be zoomed out rather than zoomed in).
I read in the manual that with matrix addition and subtraction scalars or 1x1 matrices are expanded to be conformable with the other argument, e.g.
works and results in a 3x1 matrix. Adding the next command also works as expected:
However, reversing the command results in an error:
matrix jan=jan'jan + ones(3,1)
Matrices not conformable for operation
Is this normal behaviour and am I missing something, or is this a bug?
I'm using Gretl 1.9.5cvs built 2011-04-24 on Mac OS X 10.6.8
UNIVERSITEITANTWERPEN | Faculty of Applied Economics (TEW) | Dept. Accounting & Finance
Room S.B.335 | Prinsstraat 13 | B-2000 Antwerp | Belgium
Phone +32 32654163 |Fax +32 32654064
I am Nikos, doing my thesis on real estate. I need to run 120 regressions
doing the same thing over and over again.
and save the results in files.
How can I write a script in Gretl which will do that?
I am sorry that my first post is a question and I can't contribute with
I'm a young practioner of economics and I've been trying to Seasonally
Adjust a large number of datasets related to employment.
Since I'm using a Mac, I've installed both gretl and the X-12 option. I'm
fairly tech-savvy but hit a wall so I thought I'd ask for some help.
Basically I have a dataset which includes the date, as "date", by month; and
employment, as "emplab". If I open up the gretl UX I can easily Seasonally
Adjust the data, save it to a .csv with a new variable and continue on. My
problem lies in that I need to do this frequently, and as a batch. With this
in mind I've been attempting to get X-12 to work with gretlcli in Terminal
so that I could do things as a batch that way.
My question is: Am I approaching this correctly? Is arima the correct
command? I used "help arima" in gretlcli and didn't see an option to create
a new variable for the SA data like the UX provides...
Thanks in advance,
I'd like to have some clarification on the line
"b) that the new data carries clear observation information so that
gretl can work out how to place the values"
in the description of "append" from the Gretl Command reference.
Basically, this is my situation: I have two databases. In both, I have a
two fields "id1" and "id2". The combined fields are unique in the first
database, while instead in the second I can have two different rows with
the same "id1" and "id2" values.
So, first question: when I append the second db to the first, are the
common field names "clear observation information" for Gretl to do some
sort of matching between the rows of the two databases? (as far as I can
desume from my tests, the answer is "no", but I may be missing
Second question: the second database has a field "days", and my aim is
that in my merged database, the (unique) row with given "id1" and "id2"
has, in the field "days", the sum of the "days" variable in all rows
with that combination of ids. How can I do that?
For whoever is familiar, I'm basically searching for an equivalent of
STATA's "merge" (in the m:m case) and then "compact" commands.
Thank you in advance for any hint
ols price const s
line='tabprint --format="%.4f|%.4f|%.4f|%.4f"', ci = 121
Invalid option '--format'
Note that the format line is taken straight from the helpfile. I have no
time to fix this today, I'll have a look tomorrow, if Allin doesn't
do so before then (not unlikely ;-))
Riccardo (Jack) Lucchetti
Dipartimento di Economia
Università Politecnica delle Marche