sdMedian: Estimates for standard deviations and CI for weighted medians...
In RostyslavMaiboroda/mixvconc:

Description Usage Arguments Details Value Examples

sdMedian calculates estimates of standard deviatitions and confidence intervals for weighted medians with minimax weights by observations from the mixture with varying concentrations.

1 2	sdMedian(x, p, comp = 1:ncol(p), medians = FALSE, CI = FALSE, alpha = 0.05)

`x`	numeric vector with the observed sample or a `wtsamp` object.
`p`	matrix (or data frame) of mixing probabilities with rows corresponding to subjects and columns coresponding to the mixture components.
`comp`	a numeric vector with numbers of components for which the standard deviations are estimated.
`medians`	logical, if `TRUE` then the estimates for components' medians are included in the function value.
`CI`	logical, if `TRUE` then confidence bounds for components' means are inculded in the function value. If `x` is a vector then the weights for components' medians and variances are calculated as `lsweight(p)`. If `x` is a `wtsamp` object than its own weights are used.
`alpha`	confidense level for the confidence interval.

if CI=TRUE then the function calculates confidence intervals for the components' medians with covering probability 1-alpha.

if CI & medians =FALSE the function returns a vector of the estimated standard deviations with NA for the components which were not estimated. Else a data frame is returned in which there can be variables: sd are standard deviations of estimates; medians are the estimates of medians; lower and b nupper are lower and upper bounds of the confidence intervals for medians. c

set.seed(3)
M<-3 # number of mixture components
p <- genunifp(1000,M) # create mixing probabilities
m<-c(0,1,2)      # true means of components
sd<-c(1,1,0.5)   # true sd of components
x<-genormixt(p,m,sd) # sample generation
# Calculate sd only:
sdMedian(x,p)
# the same result:
sdMedian(wtsamp(x,indiv=lsweight(p)),p)
# Calculate confidence intervals:
sdMedian(x,p,medians=TRUE,CI=TRUE)
# Plot confidence intervals:
CI<-sdMedian(x,p,medians=TRUE,CI=TRUE)
library(plotrix)
plotCI(1:M,CI$medians,ui=CI$upper,li=CI$lower,
       xlab=" ",ylab="medians",xaxt="n")
axis(1,at=1:M,labels=row.names(CI))