[R] Warnings using SuperLearner?

[Bert Gunter]

A guess:

It sounds like you are overspecifying models leading to numerical issues
during the various optimizations being performed. Try simplifying your
model (reducing the number of predictors).  Without more details on your
data and model, it may be difficult to diagnose beyond such a guess, though
there are certainy wiser folks than I out there.

Also, this is a plain text maiing list. Please post in plain text, not HTML.

Cheers,
Bert



Bert Gunter

"The trouble with having an open mind is that people keep coming along and
sticking things into it."
-- Opus (aka Berkeley Breathed in his "Bloom County" comic strip )

On Thu, Aug 9, 2018 at 7:22 AM, Kan Z. Gianattasio <
kan.gianattasio using gmail.com> wrote:

>  Hello,
>
> I'm using the CV.SuperLearner to model a binary outcome using a set of
> predictors, specifying family=binomial(), method = "method.AUC", and
> SL.library = c("SL.glmnet", "SL.glm", "SL.randomForest", "SL.gam",
> "SL.polymars", "SL.mean").
>
> I'm getting a number of warning messages (copied below), and have not been
> able to find anything about what they imply, or how to address them. Any
> insight would be greatly appreciated.
>
> Thank you in advance!
>
> ---
>
> I get the following warnings as the models are being run (many times over):
>
> step half ouch...
> step half ouch...
> warning - model size was reduced
> warning - model size was reduced
> step half ouch...
> step half ouch...
> step half ouch...
> step half ouch...
> warning - model size was reduced
> step half ouch...
> warning - model size was reduced
>
> At the very end, I'm also getting the following:
>
> 1: In method$computeCoef(Z = Z, Y = Y, libraryNames = libraryNames,  :
>   optim didn't converge when estimating the super learner coefficients,
> reason (see ?optim):  52  optim message:  ERROR:
> ABNORMAL_TERMINATION_IN_LNSRCH
> 2: In method$computeCoef(Z = Z, Y = Y, libraryNames = libraryNames,  :
>   optim didn't converge when estimating the super learner coefficients,
> reason (see ?optim):  52  optim message:  ERROR:
> ABNORMAL_TERMINATION_IN_LNSRCH
> 3: In method$computeCoef(Z = Z, Y = Y, libraryNames = libraryNames,  :
>   optim didn't converge when estimating the super learner coefficients,
> reason (see ?optim):  52  optim message:  ERROR:
> ABNORMAL_TERMINATION_IN_LNSRCH
> 4: In method$computeCoef(Z = Z, Y = Y, libraryNames = libraryNames,  :
>   optim didn't converge when estimating the super learner coefficients,
> reason (see ?optim):  52  optim message:  ERROR:
> ABNORMAL_TERMINATION_IN_LNSRCH
> 5: In method$computeCoef(Z = Z, Y = Y, libraryNames = libraryNames,  :
>   optim didn't converge when estimating the super learner coefficients,
> reason (see ?optim):  52  optim message:  ERROR:
> ABNORMAL_TERMINATION_IN_LNSRCH
> 6: In method$computeCoef(Z = Z, Y = Y, libraryNames = libraryNames,  :
>   optim didn't converge when estimating the super learner coefficients,
> reason (see ?optim):  52  optim message:  ERROR:
> ABNORMAL_TERMINATION_IN_LNSRCH
> 7: In method$computeCoef(Z = Z, Y = Y, libraryNames = libraryNames,  :
>   optim didn't converge when estimating the super learner coefficients,
> reason (see ?optim):  52  optim message:  ERROR:
> ABNORMAL_TERMINATION_IN_LNSRCH
>
>         [[alternative HTML version deleted]]
>
> ______________________________________________
> R-help using r-project.org mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/
> posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>

	[[alternative HTML version deleted]]