[Gretl-devel] Re: regls internal error
On Tue, 24 Oct 2023, Sven Schreiber wrote:
Am 23.10.2023 um 23:34 schrieb Cottrell, Allin:
> On Mon, Oct 23, 2023 at 2:39 PM Sven Schreiber
> <sven.schreiber(a)fu-berlin.de> wrote:
>> Not a big deal, but I guess nevertheless some more internal validity or
>> soundness checks are needed.
> Thanks for the report. That error indicates that sy, the return value
> from sdc() is NA, which in turn means that we had fewer than 2 valid
> observations of depvar. OK, in that case we'll call funcerr with a
> message of "Insufficient valid observations".
Thanks, Allin.
Another thing in the GUI: "randfolds" seems to be activated by default, even
though for scripting the doc says the opposite. And it seems to me that this
is especially weird in time-series datasets, if I understand the option
correctly. For time series we want to keep consecutive obs through
randfolds=FALSE, right?
Right. I'm working on this and other points relating to regls and I
expect to produce an updated version by the end of this week.
Allin