Variable selection for multilevel phylogenetic model

[guifandos]

Hi,
I am trying to use projpred for variable selection on a phylogenetic multilevel model fit with brms.
variable_response ~ x1 + x2 + (1|gr(species, cov = A))

However, I get the following error:

n model.frame.default(data = list(. variable_response.1 = c(0.580296874164707,  : 
  invalid type (list) for variable 'gr(species, cov = A)
In addition: Warning message:
Some Pareto k diagnostic values are slightly high. See help('pareto-k-diagnostic') for details.

Thanks in advance,

[fweber144]

Thanks for creating the issue. To my knowledge, special group-level terms such as (1|gr(species, cov = A)) are not supported by projpred yet. But I admit that even now, this should be documented somewhere and an appropriate error message should be thrown.

[fweber144]

But I admit that even now, this should be documented somewhere and an appropriate error message should be thrown.

Done by commits f23a8b0, 13bb505, and 7d1bda7 (directly on master).