[Gretl-devel] Re: regls and missings in the middle

Am 16.04.2025 um 14:09 schrieb Cottrell, Allin: > On Tue, Apr 15, 2025 at 10:09 AM Sven Schreiber > <sven.schreiber(a)fu-berlin.de&gt; wrote: >> Another regls (regls_multiprint) problem [...] >> >> lfrac R^2 sum(|b|) df criterion >> 1.00000 0.0000 0.000000 0 >> Uneindeutiger Matrixindex # ---> ambiguous index >> *** Fehler in Schleife in Funktion regls_multiprint >> > printf " %#10g\n", rb.crit[i] >> >> Perhaps this is a corner case? R2=0? Does somebody have a successful >> regls_multiprint invocation? > R^2 is zero by construction when lfrac = 1. Successful invocations of > multiprint are provided in the example file multiprint.inp. Please > give the context in which this error arose. Thanks, I'm attaching an example adapted from multiprint.inp which provokes the error here for me. It seems to be related to cross-validation.