Usage of pglmm()
In phyr: Model Based Phylogenetic Analysis

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>", eval = FALSE, purl = FALSE
)

The purpose of this vignette is not to provide detailed explanation of model set up and results. For this purpose, please read the documentation of ?phyr::pglmm and the chapter 4 of Ives 2018. The purpose of this vignette is to show some cases that are not documented in details.

Fit a binomial model with success and failures

To fit a binomial model to a data frame with columns of success and failures (say named as yes and no, respectively), use

pglmm(cbind(yes, no) ~ 1 + x, data = dat, family = 'binomial')

Prepare random terms for complex models that cannot be set up with formula

Sometimes, users may want to prepare their own list of random terms to fit phylogenetic generalized linear mixed models in more flexible ways. For example, users may want to add an extra random term on top of those specified by the model formula. For this, we can extract the list of random terms generated by the model formula using prep_dat_pglmm() and then append the one we want to add. See ?phyr::prep_dat_pglmm for details about its arguments.

Other functions to work with models fitted with `pglmm()`

All models fitted with pglmm() have class of communityPGLMM. Here is a list of functions that can be used to these models.

pglmm_matrix_structure(): produce the whole covariance matrix
pglmm_plot_re(): plot images of random term matrix, see vignettes plot-re
pglmm_predicted_values() or fitted(): extract fitted values
pglmm_profile_LRT(): to test significance of random terms; only works with binomial models
plot_data(): plot data used (and optionally predicted) by fitted model
plot_bayes(): plot posterior distributions of random and fixed effects
summary(): summary of model fit
print(): summary of model fit
residuals(): residuals values
fixef(): estimates of fixed effects
ranef(): estimates of random terms (variance and standard deviation)