mht.1sample {mhtboot}

Multiple hypothesis testing based on p value distribution for one sample test

Description

Implements multiple hypothesis testing based on bootstrap distribution of p values.

Usage

mht.1sample(X, B = 100, test = t.test, nbx = NROW(X), ncpus = 8,
  rbuff = 25, h = 30, qi = 0.9)

Arguments

X

matrix of data

B

bootstrap sample size, default is 100

test

one sample test. by default t.test(), user can provide own function, must return p values in $p.value

nbx

size of the bootstrap sample

ncpus

number of cpu to use

rbuff

right buffer for change detection

h

window size for change detection

qi

the quantile to use for change detection

Details

This function takes the dataset and produces the bootstrap distribtution of the transformed and ordered p values using the user given parameters. Then detects the change in the bootstrap distribution using the corner detection method. This method requires the user to specify the quantile to use for change detection. The change point is an estimate of the location of change from alternative to null and used to get the coordinates of the true signals.

Value

list with two elements. cutoff: the location of corner, signal: the index of the detected coordinates.

Examples

n = 50;m = 100;m0 = 20;
sigeff = 1;
Sigma <- 0.25*diag(m)
X <- datgen(n,m,m0,sigeff,Sigma = Sigma)
out1 <- mht.1sample(X,B=100,ncpus = 1)
out1$cutoff
out1$signal