On Fri, 21 Jul 2006, Talha Yalta wrote:
> You're making sense, if you mean to request that
> transformations of variables could be written into a model
> specification. However, I would dispute that this is a "GUI"
> feature: it requires typing the model specification. The
> gretl GUI model dialog doesn't require (or allow) any typing
> of the model spec.
Yes this is not really a GUI feature but it is indirectly
related with the GUI as well. Currently, you run a regression
and end up with a lot of newly generated variables, which
really don't have to be permanently in the variable list.
I agree, this can be an issue. It raises a larger design
When I was first setting up gretl's dialog boxes, I wondered:
How far do you have to go in making everything "mousable"? To
what extent can you just present the user with a text entry into
which they have to type something?
One of the first relevant examples I looked at was Eviews, and
there I found lots of instances where there was no
"mousability", just a text entry. At first I though, "Great, I
don't have to spend ages designing a GUI, I'll just do this sort
of thing." And the model specification dialog in early versions
of gretl was based on a simple text entry.
Since then, though, we've developed a long way in terms of
making things do-able via the mouse. (The latest substantial
step was a GUI lag selector in the model dialog, so that users
don't have to "create" lags first, before invoking them as
So far as the GUI method is concerned, I think the lag selector
is about as far as we can go. Imagine a model dialog where you
could not only select lags, but for each regressor, select
whether you wanted level, log, first difference... The
complexity becomes totally unmanageable.
Which brings us back to something like the Eviews dialog, with a
text entry, but a powerful one which enables you to embed
transformations into the list of regressors.
To some extent (OK, a small extent) we have that in the current
command-line model specification. For example:
ols y 0 x time
ols y 0 x x(-1)
ols y 0 x(-1 to -3)
ols y 0 dummify(z)
where the trend, "time", the lags of x, and the dummy variables
representing the values of the (discrete) variable z, did not
pre-exist the command.
It would not be very difficult to generalize this, and allow
arbitrary transformations of variables to be specified as
regressors. But, in the very nature of things, there's no real
GUI for this sort of functionality.
Is it, then, worth considering a bifurcation of the model
specification dialog? A "user-friendly" option (which we have
already) and an "expert" version (which sticks a vamped-up
version of the current command-line model specification into a
text-entry dialog box)?