Package index • mcmcglm

Sampling posterior distributions of parameters in a GLM

Function to perform sampling as well as methods to investigate results

mcmcglm(): Efficient Gibbs sampling of posterior distribution of parameters in GLM

Methods for `mcmcglm` objects

Methods to explore results - data of samples and function for creating trace plot

coef(<mcmcglm>): S3 method for getting the average value of coefficients

quantile(<mcmcglm>): S3 method for getting quantiles of samples for each parameter

samples(): Get the drawn samples from the object

trace_plot(): Create a trace plot of the MCMC samples

Tuning parameters of slice sampling

Functions for getting a list of results across tuning parameter values and plotting them

mcmcglm_across_tuningparams(): Get list of mcmcglms run across values of slice sampling tuning parameters

plot_mcmcglm_across_tuningparams(): Plot a list of mcmcglms showing varying tuning parameters in the title of the plots

Tools for investigating computation time for different methods

A function which returns a data frame of computation time for different specifications and method for plotting results

compare_eta_comptime_across_nvars(): Compare runtime using CGGibbs and naive approach to calculate linear predictor

plot_eta_comptime(): Plot the results of compare_eta_comptime_across_nvars

Calculating the log-potential

Utilities for calculating the log-potential as a function of a new coordinate of parameter vector in order to perform slice sampling within Gibbs

log_density(): S3 generic for calculating the log density of a distribution dispatched via a family

log_likelihood(): Calculate log likelihood parametrised by "mu"

update_linear_predictor(): Update value of a linear predictor as function of a single coordinate change

log_potential_from_betaj(): Calculate the log-potential (log-likelihood plus log-density of prior)