plotCurveEstimate: Plot function value estimates

In hypergsplines: Bayesian model selection with penalised splines and hyper-g prior

Description

Plot function value estimates

Usage

  plotCurveEstimate(covName, samples, modelData,
    nGrid = 200L, plevel = 0.95, slevel = plevel,
    partialResids = TRUE, hpd = TRUE,
    estimate = c("mean", "median"), plot = TRUE, ...,
    ylim = NULL, lty = c(1, 2, 2, 2, 2),
    col = c("black", "blue", "blue", "green", "green"),
    xlab = covName,
    ylab = paste("f(", covName, ")", sep = ""))

Arguments

covName	string with the name of the covariate
samples	the samples object (either from getSamples or the samples element from glmGetSamples)
modelData	the corresponding model data object
nGrid	number of abscissa values for the grid (default: 200)
plevel	credible level for the pointwise credible intervals (default: 0.95, and NULL suppresses it)
slevel	credible level for simultaneous credible band (defaults to plevel, and NULL suppresses it)
partialResids	add partial residuals to the plot? (default, is only possible for normal models)
hpd	use HPD intervals / bands? (default) Otherwise equi-tailed intervals / bands are computed and plotted.
plot	produce a plot? (default: yes) Otherwise you could use the returned information to produce the plot yourself
estimate	type of the estimated curve
ylim	y axis limits (has a sensible default to include all points in the plot)
lty	line type for (1) mean curve and (2, 3) credible interval bounds (default: c(1, 2, 2, 2, 2))
col	line color(s) (default: black, blue, blue, green, green)
xlab	x axis label (default: covName)
ylab	y axis label (default: f(covName) )
...	... additional plotting parameters

Value

A list with the elements x.grid and lineData, containing the x axis grid and the different y values for plotting, respectively.

Author(s)

Daniel Sabanes Bove [e-mail]

Examples

## get some data
attach(longley)

## get model data
md <- modelData(y=Employed,
                X=cbind(GNP, Armed.Forces),
                gPrior="hyper-g")

## get posterior samples for a specific model configuration
samples <- getSamples(config=c(2, 1),
                      nSamples=1000L,
                      modelData=md)
str(samples)
summary(samples$t)

## and plot resulting curve estimates:
par(mfrow=c(1, 2))

plotCurveEstimate(covName="GNP",
                  samples=samples,
                  modelData=md)

plotCurveEstimate(covName="Armed.Forces",
                  samples=samples,
                  modelData=md)

plotCurveEstimate(covName="Armed.Forces",
                  samples=samples,
                  modelData=md,
                  hpd=FALSE,
                  estimate="median")

hypergsplines documentation built on May 2, 2019, 6:14 p.m.