Bayz is set-up as other R model functions to produce an output object that can be used for making summaries, plots, retrieve estimates, and many other things. The standard use is the following workflow that will be familiar to R users:
library(BayzR) fit = bayz(y~ ...) summary(fit)Bayz now supports the following R generic functions to work on the model output: summary(), plot(), coef(), fixef(), ranef(), predict(). Below is a description of these functions; standard help files for each are also available in R.
The bayz summary() reports: 1) a table of Estimates of model coefficients (estimates of fixed and random effects); 2) a table with convergence diagnostics and HPD intervals for 'logged' parameters; 3) a table with proportional (scaled) variance estimates, including HPD intervals.
The bayz plot() function produces trace plots for the 'logged' parameters.
The coef(), fixef() and ranef() functions supply ways to retrieve model estimates of fixed and random effects from the bayz output object. Coef() retrieves all coefficient estimates in a list with one data-frame for each model-term fitted. The list elements are names as the original variables fitted in the model-terms. Fixef() and ranef() are specialized versions of coef() that only retrieve the fixed effects and random effects respectively.
The predict() function in bayz is set-up to provide predictions for NA values that were in the original data-frame used to fit the model. This allows standard cross-validation work-flows to set parts of data to NA and run the model to obtain prediction.