there's a feature requests on logit models that is now one year old, see
[ 1886339 ]:
Of course, in principle there's no reason why a feature request couldn't
simply exist for years until it might eventually get fulfilled; but
still I think one year is a good time to make a decision whether it is
still needed in the feature request database.
For example, the first part of that request (implement multinomial
logit) is of course valid, but in some sense it's not needed, because
everybody knows that having more models in gretl would be nice, it's
just a matter of limited programmer resources, and a reminder isn't
necessary. For the same reason I haven't filed a feature request
"implement structural VAR models" or similar things.
So if it were just for that part of the request, I would be inclined to
But then there are the smaller parts 2 and 3 which relate to the
handling of the reference category. What about those, is this deemed
unimportant or could it be done in the next half-year or so? And BTW,
wouldn't this refer to probit also? (and to what else?) So maybe we
should rename the request along those lines?