bartMachine (version 1.2.6)

plot_convergence_diagnostics: Plot Convergence Diagnostics

Description

A suite of plots to assess convergence diagonstics and features of the BART model.

Usage

plot_convergence_diagnostics(bart_machine,  
plots = c("sigsqs", "mh_acceptance", "num_nodes", "tree_depths"))

Value

None.

Arguments

bart_machine

An object of class ``bartMachine''.

plots

The list of plots to be displayed. The four options are: "sigsqs", "mh_acceptance", "num_nodes", "tree_depths".

Author

Adam Kapelner and Justin Bleich

Details

The ``sigsqs'' option plots the posterior error variance estimates by the Gibbs sample number. This is a standard tool to assess convergence of MCMC algorithms. This option is not applicable to classification BART models.
The ``mh_acceptance'' option plots the proportion of Metropolis-Hastings steps accepted for each Gibbs sample (number accepted divided by number of trees).
The ``num_nodes'' option plots the average number of nodes across each tree in the sum-of-trees model by the Gibbs sample number (for post burn-in only). The blue line is the average number of nodes over all trees.
The ``tree_depths'' option plots the average tree depth across each tree in the sum-of-trees model by the Gibbs sample number (for post burn-in only). The blue line is the average number of nodes over all trees.

Examples

Run this code
if (FALSE) {
#generate Friedman data
set.seed(11)
n  = 200 
p = 5
X = data.frame(matrix(runif(n * p), ncol = p))
y = 10 * sin(pi* X[ ,1] * X[,2]) +20 * (X[,3] -.5)^2 + 10 * X[ ,4] + 5 * X[,5] + rnorm(n)

##build BART regression model
bart_machine = bartMachine(X, y)

#plot convergence diagnostics
plot_convergence_diagnostics(bart_machine)
}

Run the code above in your browser using DataCamp Workspace