New gretl4py Release – Version 0.50
by Marcin Błażejowski
Hi,
I’m excited to announce a new release of the gretl4py Python module!
Version 0.50 includes significant updates, upgrades, and bug fixes. The
highlights are:
*
Several bugs fixed (including the one related to exporting nested
bundles to a dictionary)
*
Extensive documentation updates
*
General |libgretl| printer now available as |gretl.print()|
*
Code refactoring for improved clarity and simplicity
*
Official interface to |libgretl|’s |genr| added (|gretl.genr()|)
*
Robustified |install.py| (thanks, Allin!)
*
New test: |fractint|
*
New |_gretl.Dataset| methods: |::hurst()|, |::xcorrgm()|, |::pca()|,
|::rename()|
*
New free functions: |corr()|, |cov()|, |pkg()|
*
New |GretlPackage| methods: |::get_functions()|, |::get_help()|
*
|GretlBundle::pop()| method added + enhanced |[]| operator
*
New |GretlBundle| constructor for on-the-fly bundle creation
(similar to gretl’s |_()| function)
*
|get_data()| now supports a |quiet| option
*
Marked |nls|, |mle|, |gmm|, and system estimators as experimental
*
Refactored |GretlModel::fcast()| to mimic libgretl’s |fcast| command
*
New |GretlModel::dataset| attribute points directly to the dataset
used for model estimation
*
|_gretl.Dataset::varnames()| method added as |::varnames| attribute
*
New |_gretl.Dataset::summary()| method
*
|GretlPackage| examples added for BACE, BayTool, BMA, BVAR,
criteria, gig, ParMA, SVAR, and TVC
*
|GretlPackage| parameters now accept both positional (|*args|) and
keyword (|**kwargs|) arguments
Have fun exploring the new features!
Best regards,
Marcin
1 week, 3 days
about "Turning observation-marker strings into a series"
by Sven Schreiber
Hi,
the cheatsheet chapter in the gretl user guide has a part about this
subject. Apart from using the "markers" command, it proposes to use the
stringify function in a second step.
However, in section 16.2 of the same guide, under the heading of
"Assigning from an array of strings" there's an example that also starts
with the markers command, but then simply assigns the result to a
series, as in "series state = S". I just verified in my own panel
dataset that that worked nicely. (Full disclosure: My dataset only had
the country codes in the observation markers, no panel-time information
in there. That would have complicated things, I guess., But I think the
same problem would apply to the first solution.)
So the question is: Given that the second way feels more natural, is the
solution in the cheatsheet outdated, or is it better in some other
respect that I'm missing?
thanks
sven
1 week, 5 days
Package updates (January 2026)
by Riccardo (Jack) Lucchetti
Dear all,
this message is to inform the community about the activity in our
function package repository: during the month of january 2025, 10
packages were updated to a new version and a new one was released.
The new one is
"DYconn", by C. Casoli, L. Pedini and A. Tarassow: it implements Diebold
and Yilmaz' VAR-based connectedness measures. If you know what I'm
talking about, good. If you don't , you may want to have a look at
Chiara and Luca's nice paper on the subject
<https://link.springer.com/article/10.1007/s00180-025-01680-9>.
The updates were:
"heatmap", by Artur Tarassow (heatmap, contour and 3D plots)
"FEP", by Artur Tarassow and Sven Schreiber (Forecast Evaluation via
many measures)
"yahoo_get", by myself (Downloader of daily financial data from Yahoo)
"ARMA_IRF", again by myself (IRF for univariate ARMA models)
"gnuplot_piechart" , by Ignacio Díaz-Emparanza (A simple 2D pie chart)
"ketvals", by Francesco Valentini and myself (Kernel-based Time-Varying
Least Squares)
"StrucTiSM", by me and Sven Schreiber (Harvey-style Structural Time
Series Models)
"adalasso", also by Sven Schreiber (adaptive Lasso)
"DynMultCalc", by Allin and me (dynamic multipliers for ADL models)
"ARprobit", again by Sven Schreiber (Dynamic Probit à la Kauppi&Saikkonen)
Yes, there's quite a few of them. Several, however, contain only
cosmetic adjustments, needed to silence a warning message about
deprecated syntax for drawing bands in plots. That said, some others
contain nice substantial improvements: for example, "FEP" has some nice
improvements in the Giacomini-White test, "StrucTiSM" finally includes
auxiliary residuals (a long-awaited feature) and a nasty bug in
"adalasso" was fixed.
As usual, update and enjoy!
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
2 weeks, 2 days
gretl 2026a for Windows
by Allin Cottrell
I'm sorry to say that we seem to have missed a few required DLLs
for some of the builds of gretl 2026a for Windows. Specifically:
* gretl-2026a-64.exe and gretl-2026a-win64.zip (both for Intel/AMD
computers) were missing a DLL required for the command-line program,
gretlcli.exe.
* gretl-2026a-arm64.exe was missing a DLL required for the GUI
program, gretl.exe.
This is now fixed in new builds, dated 2026-01-31, at
https://sourceforge.net/projects/gretl/files/gretl/2026a/
Allin Cottrell
2 weeks, 5 days