[Gretl-devel] Re: RE probit and perfect prediction
On Sat, 7 Nov 2020, Riccardo (Jack) Lucchetti wrote:
It's very unlikely that I'll have time to work on this over
the weekend, but
will try to do so on Monday. If you have alternative ideas on how to handle
this, suggestions are welcome!
Replying to myself: ths fix was much much easier that I had thought:
diff --git a/plugin/reprobit.c b/plugin/reprobit.c
index 9b6c965d3..2b42b6a7d 100644
--- a/plugin/reprobit.c
+++ b/plugin/reprobit.c
@@ -665,7 +665,6 @@ MODEL reprobit_estimate (const int *list, DATASET *dset,
} else if (mod.list[0] < list[0]) {
/* something was dropped */
gretl_errmsg_set("Couldn't estimate initial probit");
- err = E_DATA;
}
if (!err) {
I wonder if it could really be as easy as this. How come we were flagging
an error in that case? It feels as if I'm missing something. Allin, do you
remember why we were setting err to E_DATA?
-------------------------------------------------------
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
-------------------------------------------------------