svyhist {survey} | R Documentation |
Histograms and boxplots weighted by the sampling weights.
svyhist(formula, design, breaks = "Sturges",
include.lowest = TRUE, right = TRUE, xlab = NULL,
main = NULL, probability = TRUE, freq = !probability, ...)
svyboxplot(formula, design, all.outliers=FALSE,...)
formula |
One-sided formula for |
design |
A survey design object |
xlab |
x-axis label |
main |
Main title |
probability,freq |
Y-axis is probability density or frequency |
all.outliers |
Show all outliers in the boxplot, not just extremes |
breaks, include.lowest, right |
As for |
... |
The histogram breakpoints are computed as if the sample were a simple random sample of the same size.
The grouping variable in svyboxplot
, if present, must be a factor.
The boxplot whiskers go to the maximum and minimum observations or to
1.5 interquartile ranges beyond the end of the box, whichever is
closer. The maximum and minimum are plotted as outliers if they are
beyond the ends of the whiskers, but other outlying points are not
plotted unless all.outliers=TRUE
. svyboxplot
requires a two-sided formula; use variable~1
for a single boxplot.
As for hist
, except that when probability=FALSE
, the return value includes a component
count_scale
giving a scale factor between density and
counts, assuming equal bin widths.
data(api)
dstrat <- svydesign(id = ~1, strata = ~stype, weights = ~pw, data = apistrat,
fpc = ~fpc)
opar<-par(mfrow=c(1,3))
svyhist(~enroll, dstrat, main="Survey weighted",col="purple",ylim=c(0,1.3e-3))
hist(apistrat$enroll, main="Sample unweighted",col="purple",prob=TRUE,ylim=c(0,1.3e-3))
hist(apipop$enroll, main="Population",col="purple",prob=TRUE,ylim=c(0,1.3e-3))
par(mfrow=c(1,1))
svyboxplot(enroll~stype,dstrat,all.outliers=TRUE)
svyboxplot(enroll~1,dstrat)
par(opar)