bartMachine: plot_y_vs_yhat – R documentation

Become an expert in R — Interactive courses, Cheat Sheets, certificates and more!

plot_y_vs_yhat

Plot the fitted Versus Actual Response

Description

Generates a plot actual versus fitted values and corresponding credible intervals or prediction intervals for the fitted values.

Usage

plot_y_vs_yhat(bart_machine, Xtest = NULL, ytest = NULL, 
credible_intervals = FALSE, prediction_intervals = FALSE, 
interval_confidence_level = 0.95)

Arguments

`bart_machine`	An object of class “bartMachine”.
`Xtest`	Optional argument for test data. If included, BART computes fitted values at the rows of `Xtest`. Else, the fitted values from the training data are used.
`ytest`	Optional argument for test data. Vector of observed values corresponding to the rows of `Xtest` to be plotted against the predictions for the rows of `Xtest`.
`credible_intervals`	If TRUE, Bayesian credible intervals are computed using the quantiles of the posterior distribution of \hat{f}(x). See `calc_credible_intervals` for details.
`prediction_intervals`	If TRUE, Bayesian predictive intervals are computed using the a draw of from \hat{f}(x). See `calc_prediction_intervals` for details.
`interval_confidence_level`	Desired level of confidence for credible or prediction intervals.

Value

None.

Note

This function is parallelized by the number of cores set in set_bart_machine_num_cores.

Author(s)

Adam Kapelner and Justin Bleich

Examples

## Not run: 
#generate linear data
set.seed(11)
n  = 500 
p = 3
X = data.frame(matrix(runif(n * p), ncol = p))
y = 3*X[ ,1] + 2*X[ ,2] +X[ ,3] + rnorm(n)

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

##generate plot
plot_y_vs_yhat(bart_machine)

#generate plot with prediction bands
plot_y_vs_yhat(bart_machine, prediction_intervals = TRUE)

## End(Not run)

bartMachine

Bayesian Additive Regression Trees

v1.2.6

GPL-3

Authors

Adam Kapelner and Justin Bleich (R package)

Initial release

2020-12-17