bartMachine - BART

This guide is designed as a quick-stop reference of how to use some of the more popular machine learning R packages with vivid. In the following examples, we use the air quality data for regression and the iris data for classification.

bartMachine - BART

The bartMachine package in R provides an implementation of Bayesian Additive Regression Trees (BART), a non-parametric Bayesian model which uses decision trees as the underlying model. To begin we load the packages and data.

library('vivid')
library("bartMachine")

Regression

# load data 
aq <- na.omit(airquality)


# build bartMachine model
bm <- bartMachine(X = aq[,2:6], 
                  y = aq[,1], 
                  num_trees = 20,
                  num_burn_in = 100,
                  num_iterations_after_burn_in = 500)


# vivi matrix
vi <- vivi(data = aq, fit = bm, response = 'Ozone')

Heatmap

viviHeatmap(mat = vi)

Figure 1: Heatmap of a BART regression fit displaying 2-way interaction strength on the off diagonal and individual variable importance on the diagonal.

PDP

pdpPairs(data = aq, 
         fit =  bm, 
         response = "Ozone", 
         nmax = 500, 
         gridSize = 10,         
         nIce = 50)

Figure 2: Generalized pairs partial dependence plot for a BART regression fit.

Classification

# Get data and only use 2 factors
data(iris)
iris2 = iris[51:150,]
iris2$Species = factor(iris2$Species)

bm <- build_bart_machine(iris2[ ,1:4], iris2$Species,
                         num_trees = 20,
                         num_burn_in = 100,
                         num_iterations_after_burn_in = 500)

# vivid
vi <- vivi(data = iris2, fit = bm, response = 'Species')

Heatmap

viviHeatmap(mat = vi)

Figure 3: Heatmap of a BART classification fit displaying 2-way interaction strength on the off diagonal and individual variable importance on the diagonal.

PDP

pdpPairs(data = iris2, 
         fit =  bm, 
         response = "Species", 
         nmax = 500, 
         gridSize = 10,         
         nIce = 50)

Figure 4: Generalized pairs partial dependence plot for a BART classification fit.