| svycdf {survey} | R Documentation |
Estimates the population cumulative distribution function for specified
variables. In contrast to svyquantile, this does not do
any interpolation: the result is a right-continuous step function.
svycdf(formula, design, na.rm = TRUE,...) ## S3 method for class 'svycdf': print(x,...) ## S3 method for class 'svycdf': plot(x,xlab=NULL,...)
formula |
one-sided formula giving variables from the design object |
design |
survey design object |
na.rm |
remove missing data (case-wise deletion)? |
... |
other arguments to plot.stepfun |
x |
object of class svycdf |
xlab |
a vector of x-axis labels or NULL for the default labels |
An object of class svycdf, which is a list of step functions (of
class stepfun)
svyquantile, svyhist, plot.stepfun
data(api)
dstrat <- svydesign(id = ~1, strata = ~stype, weights = ~pw, data = apistrat,
fpc = ~fpc)
cdf.est<-svycdf(~enroll+api00+api99, dstrat)
cdf.est
## function
cdf.est[[1]]
## evaluate the function
cdf.est[[1]](800)
cdf.est[[2]](800)
## compare to population and sample CDFs.
opar<-par(mfrow=c(2,1))
cdf.pop<-ecdf(apipop$enroll)
cdf.samp<-ecdf(apistrat$enroll)
plot(cdf.pop,main="Population vs sample", xlab="Enrollment")
lines(cdf.samp,col.points="red")
plot(cdf.pop, main="Population vs estimate", xlab="Enrollment")
lines(cdf.est[[1]],col.points="red")
par(opar)