mmctest-methods {simctest}

Methods for class ‘mmctest’, Package ‘simctest’

Description

Constructor for class ‘mmctest’.

Usage

 mmctest(epsilon=0.01, threshold=0.1, r=10000, h, thompson=F, R=1000)

Arguments

epsilon

probability of any misclassification one is willing to tolerate

threshold

threshold for testing.

r

parameter of the spending sequence, see vignette

h

reference to a multiple testing function of the form function(p, threshold) which returns the set of rejected indices.

thompson

if set to true, mmctest will use a Thompson strategy to draw further samples

R

number of repetitions (=draws from the posterior distributions) used to calculate empirical probabilities of each hypothesis being rejected – used to calculate weights in QuickMMCTest (option thompson=TRUE in the mmctest constructor)

Methods

mmctest(epsilon=0.01, threshold=0.1, r=10000, h)

returns object of type ‘mmctest’.

Examples

  fun <- function(ind,n,data) sapply(1:length(ind), function(i) sum(runif(n[i])<=data[ind[i]]));
  i <- mmctSampler(fun,num=500,data=runif(500));
  a <- mmctest(h=hBH);
  a <- run(a, i, maxsteps=list(maxnum=1000000,undecided=10));