plotTrace {plotMCMC} | R Documentation |
Plot MCMC Traces
Description
Plot Markov chain Monte Carlo traces. This is a diagnostic plot for deciding whether a chain shows unwanted trends.
Usage
plotTrace(mcmc, axes=FALSE, same.limits=FALSE,
between=list(x=axes,y=axes), div=1, span=1/4, log=FALSE,
base=10, main=NULL, xlab=NULL, ylab=NULL, cex.main=1.2,
cex.lab=1, cex.axis=0.8, cex.strip=0.8, col.strip="gray95",
las=0, tck=0.5, tick.number=5, lty.trace=1, lwd.trace=1,
col.trace="gray", lty.median=1, lwd.median=1,
col.median="black", lty.loess=2, lwd.loess=1,
col.loess="black", plot=TRUE, ...)
Arguments
mcmc |
MCMC chain(s) as a vector, data frame or |
axes |
whether axis values should be plotted. |
same.limits |
whether panels should have same x-axis limits. |
between |
list with |
div |
denominator to shorten values on the y axis. |
span |
smoothness parameter, passed to |
log |
whether values should be log-transformed. |
base |
logarithm base. |
main |
main title. |
xlab |
x-axis title. |
ylab |
y-axis title. |
cex.main |
size of main title. |
cex.lab |
size of axis labels. |
cex.axis |
size of tick labels. |
cex.strip |
size of strip labels. |
col.strip |
color of strip labels. |
las |
orientation of tick labels: 0=parallel, 1=horizontal, 2=perpendicular, 3=vertical. |
tck |
tick mark length. |
tick.number |
number of tick marks. |
lty.trace |
line type of trace. |
lwd.trace |
line width of trace. |
col.trace |
color of trace. |
lty.median |
line type of median. |
lwd.median |
line width of median. |
col.median |
color of median. |
lty.loess |
line type of loess. |
lwd.loess |
line width of loess. |
col.loess |
color of loess. |
plot |
whether to draw plot. |
... |
passed to |
Value
When plot=TRUE
, a trellis plot is drawn and a data frame is
returned, containing the data used for plotting. When
plot=FALSE
, a trellis object is returned.
Note
The Args
function from the gdata package is recommended
for reviewing the arguments, instead of args
.
See Also
xyplot
and panel.loess
are the underlying drawing functions, and
traceplot
is a similar non-trellis plot.
plotTrace
, plotAuto
, plotCumu
, and
plotSplom
are diagnostic plots.
plotDens
and plotQuant
are posterior
plots.
plotMCMC-package
gives an overview of the package.
Examples
plotTrace(xpar, xlab="Iterations", ylab="Parameter value",
layout=c(2,4))
plotTrace(xpar$R0, axes=TRUE, div=1000)