Quor (version 0.1)

conf.statement: Confidence Statement for Ordered Population Quantiles

Description

Evaluates the Confidence Statement for Ordered Population Quantiles.

Usage

conf.statement(data,quantiles=NULL,ordering=NULL,verbose=TRUE,logscale=FALSE) conf.statement.pooled(data,quantiles=NULL,ordering=NULL, verbose=TRUE,logscale=FALSE)

Arguments

data

is a list with all groups to be test. Each element can be a vector with the elements, or a matrix, in which case each row will be considered as a different covariate to be tested. All the elements in the list must have the exact same dimension.

quantiles

a vector of elements in [0,1] with length equal to the number of groups. It tells us which quantile will be used for each group. If null, then medians are compared by default.

ordering

a matrix containing one permutation of 1:n per row, where n is the number of groups. It tells us which orderings are to be tested. If null, then it is assumed all orderings to be tested.

verbose

if TRUE, display results on screen, FALSE otherwise.

logscale

if TRUE, return confidence values in log scale.

Value

A list (with class "conf.statement") with components:

call

command evaluated.

confidence

the (log) confidence statement level as a matrix permutations by variables.

quantiles

the quantiles of interested for each of the k groups.

order

the permutations of groups that were used.

total.groups

the total number of groups.

total.covariates

the total number of variables on which the confidence was computed.

run.time

Total time spent to run it.

Details

The confidence statement is evaluated for the population quantiles consedering: The population median of the group 1 is lower than the population median of the group 2, and this is lower than the population median of the group 3, and so... That is, M1 < M2 < ... < Mk, where Mj is the population median of the group j.

The group j is the j-th vector in the data (a list object).

If the conf.statement.pooled version is used, then each groups is checked against all the others, which are pooled together, so the computation always involves only two groups at a time (the target group and the pooled one).

Examples

  set.seed(42)
  # Statement: Q1 < Q2 < Q3, where Qj is the population
  #   median of the group j.
  data <- NULL
  data$x1 <- rnorm(10,0,1)
  data$x2 <- rnorm(10,1,1)
  data$x3 <- rnorm(10,2,1)
  conf.statement(data,ordering=as.integer(c(1,2,3)),verbose=FALSE)

  # Statement: Q2 < Q3 < Q1, where Qj is the population
  #   3rd quartile of the group j.
  conf.statement(data,ordering=as.integer(c(2,3,1)),
                 quantiles=c(0.75,0.75,0.75))

  # Evaluates statements Q1 < Q2 and Q2 < Q1 for the
  # data set gleason7, where Qj is the median of
  # population j
  data(gleason7)
  d <- list(x1 = gleason7[1:5,1], x2 = gleason7[,2])
  conf.statement(d,verbose=FALSE)