rbartonsimd: rbart run on simulated data

In rbart: Bayesian Trees for Conditional Mean and Variance

Description

predict.rbart results for simulated data.

Usage

data("rbartonsimd")

Format

rbartonsimd

returned list from a call to predict.rbart on the simulated data.

Details

The data rbartonsimd is the results of an rbart run on the simulated data in simdat.

The code for the rbart run is:

## load data
data(simdat)
attach(simdat) #some people think this is a bad idea

## run rbart
set.seed(99)
res = rbart(x,y,nskip=500,ndpost=100,nadapt=500,adaptevery=50)
rbartonsimd = predict(res,xp) #get prediction for test x in xp

Examples

## load simulated data and rbart run.
data(rbartonsimd)
data(simdat)

## plot data and x vs f(x), x vs f(x) +/- 2s(x), x in test data, true and estimated
## data
plot(simdat$xp,simdat$yp)
## true
lines(simdat$xp,simdat$fxp,col="blue",lty=2,lwd=2)
lines(simdat$xp,simdat$fxp+2*simdat$sxp,col="blue",lty=2,lwd=2)
lines(simdat$xp,simdat$fxp-2*simdat$sxp,col="blue",lty=2,lwd=2)
##estimated
mhat = rbartonsimd$mmean; shat = rbartonsimd$smean
lines(simdat$xp,mhat,col="red",lty=1,lwd=2)
lines(simdat$xp,mhat + 2*shat,col="red",lty=1,lwd=2)
lines(simdat$xp,mhat - 2*shat,col="red",lty=1,lwd=2)

## note that you can get "nicer" looking fits by 
## (i) running rbart longer (e.g. ndpost=500), 
## (ii) using numcut=1000,k=5 in rbart.

rbart documentation built on Aug. 1, 2019, 5:04 p.m.