Implement categorical predictors

[abraao-crypto]

Hi,

Firstly, I would like to congrats you on this powerful package.
I am using the function dSep(x,FUN=MCMCglmm,...) in order to do a path analysis, but unfortunately, I am having this error:
Error in parse(text = x, keep.source = FALSE) :
:2:0: unexpected end of input
1: NULL ~
^

Please, can you help me?

All best/
Abraão

[jmaspons]

Thanks for your kind words. Are there any categorical variable in your model? Can you provide a minimum example to reproduce the error?

[abraao-crypto]

Hi.
I attached a script, phylogeny, and one dataset.

dSep_Example.zip

Thanks

[abraao-crypto]

Yes , there is only one categorical variable in my model.

Thanks

[jmaspons]

Ok, handling of categorical variables is on the TODO list and the implementation shouldn't be problematic (see https://github.com/jmaspons/dSep/blob/00b899cdbde2aeea14a25780b6c446b4c31a209f/R/D-separation.R#L298), but there is some methodological questions to answer: which p-value to take from the different levels of the categorical predictors? I don't have the answer but if you have references about how to do it, I will happily implement it.

[jmaspons]

Perhaps @AlejandroG-V have some idea about how to handle categorical variables in the d-separation method...

[abraao-crypto]

Ok thanks, I think p-value < 0.05. Thanks Em seg, 10 de mai de 2021 07:28, jmaspons ***@***.***> escreveu:

…

Ok, handling of categorical variables is on the TODO list and the implementation shouldn't be problematic (see https://github.com/jmaspons/dSep/blob/00b899cdbde2aeea14a25780b6c446b4c31a209f/R/D-separation.R#L298 ), but there is some methodological questions to answer: which p-value to take from the different levels of the categorical predictors? I don't have the answer but if you have references about how to do it, I will happily implement it. Missatge de abraao-crypto ***@***.***> del dia dj., 6 de maig 2021 a les 0:08: > Yes , there is only one categorical variable in my model. > > Thanks > > — > You are receiving this because you commented. > Reply to this email directly, view it on GitHub > <#1 (comment)>, or > unsubscribe > < https://github.com/notifications/unsubscribe-auth/AAAZB5EJVBVPE7WZ6UR7LIDTMHMWRANCNFSM4364MZJA > > . > -- Salut! Joan — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#1 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/APQSJCF2ZIWMF5LS5XPC5ETTM6YNXANCNFSM4364MZJA> .

[jmaspons]

We need only one p-value and for categorical predictors we get a p-value for each level of the variable - 1 (the reference level has no p-value)