[R-sig-ME] Is there a way to deal with errors such as this?

D. Rizopoulos d@r|zopou|o@ @end|ng |rom er@@mu@mc@n|
Sun Dec 15 22:08:52 CET 2019

In the datasets for which glmer() fails you could try fitting the model with GLMMadaptive or glmmTMB. The model you’re fitting is always the same, the optimization and numerical integration algorithms change in the different packages.

Moreover, if the main aim of the simulation is to assess some properties of the model and not of the optimization algorithm, you could help the optimization procedure by supplying as starting values for the model parameters the true parameters values from which you simulate the data.


Dimitris Rizopoulos
Professor of Biostatistics
Erasmus University Medical Center
The Netherlands

Από: Ο χρήστης R-sig-mixed-models <r-sig-mixed-models-bounces using r-project.org> εκ μέρους του χρήστη Rolf Turner <r.turner using auckland.ac.nz>
Στάλθηκε: Κυριακή, Δεκεμβρίου 15, 2019 21:45
Προς: Daniel Lüdecke
Κοιν.: r-sig-mixed-models using r-project.org
Θέμα: Re: [R-sig-ME] Is there a way to deal with errors such as this?

On 15/12/19 11:20 pm, Daniel Lüdecke wrote:

> The model seems to fit w/o error when you use "glmmTMB". Unlike
> glmmADAPTIVE, which uses a more time consuming adaptive Gaussian
> quadrature rule, glmmTMB might be faster to run the models (even faster
> than glmer(), probably).


Thanks for this advice. For some reason I had it my head that
GLMMadaptive worked better than glmmTMB. I don't know why I got that
idea. I will do some experimentation and timing assessments later on.
But for the moment I'm sticking with glmer() and the strategy of
discarding the simulated data set and generating a new one if glmer()
throws an error.

Thanks again.



Honorary Research Fellow
Department of Statistics
University of Auckland
Phone: +64-9-373-7599 ext. 88276

R-sig-mixed-models using r-project.org mailing list

	[[alternative HTML version deleted]]

More information about the R-sig-mixed-models mailing list