data
argument is a bigdata bdFrame (see the DETAILS).
qqmath(formula, distribution=qnorm, f.value=ppoints, ...)The following arguments have special meaning within this function. The common meanings for these and all other arguments are listed separately under
trellis.args.
~ x | g1 * g2 * ...
however the given variables
g1,
g2 ,
... may be omitted.
x is a numeric vector.
qnorm,
qunif , or
your own function (see the examples).
function(n) c(.01, .05, seq(.1,.9,.1), .95, .99)
trellis,
which is automatically plotted by
print.trellis.
If the
data argument is a bdFrame then approximate quantiles
(based on binning the data into 1000 equi-spaced bins)
are computed and plotted.
# how well is a normal sample fit by a t distn on 7 df?
qqmath( ~ rnorm(100), distribution=function(p) qt(p, df=7))
qqmath( ~ height | voice.part, data=singer,
prepanel = prepanel.qqmathline,
panel = function(x, y, ...) {
panel.grid()
panel.qqmathline(y, distribution=qnorm, ...)
panel.qqmath(x, y, ...)
},
layout=c(2, 4), aspect=1,
xlab="Unit Normal Quantile", ylab="Height (inches)")
# residuals for each voice part plotted against distn of pooled residuals
qqmath( ~ height | voice.part, data=as.bdFrame(singer),
prepanel = prepanel.qqmathline,
panel = function(x, y, ...) {
panel.grid()
panel.bd.qqmathline(x, y, ...)
panel.qqmath(x, y, ...)
},
layout=c(2, 4), aspect=1,
xlab="Unit Normal Quantile", ylab="Height (inches)")
# like above but for bigdata. Note x argument to panel.bd.qqmathline
# and the lack of distribution=qnorm argument.
attach(singer)
oneway.residuals <- oneway(height ~ voice.part, spread=1)$residuals
qqmath( ~ oneway.residuals | voice.part,
distribution=function(p) quantile(oneway.residuals, p),
panel=function(x, y) {
panel.grid()
panel.abline(0, 1)
panel.qqmath(x, y)
},
aspect=1, layout=c(2, 4),
xlab="Pooled Residual Height (inches)",
ylab="Residual Height (inches)")
# specifying different plotting parameters in the panel function
qqmath(~Weight|Fuel, panel=function(...) panel.qqmath(..., cex= 2))