wj2.permuted: The null distribution

In andrewthomasjones/EMMIXcontrasts: Contrasts in Mixed Effects for EMMIX Model with Random Effects

Description

This function caculates the null distribution of the weighted contrast W_j.

Usage

wj2.permuted(data, ret, nB = 99, contrast = NULL, seed = 1234)

Arguments

data	The dataset, an n by m numeric matrix, where n is number of observations and m the dimension of data
ret	The return list of function emmixwire
nB	The number of permutations
contrast	A two- or three- dimensional vector the contrast(s) for the class differences
seed	random seed for the permutations.

Details

The number of classes of samples is either two or three, and the default contrast for two classes is c(1, -1), and three classes c(1, 0, -1).

Value

An n by nB matrix with its columns as the statistic Wj for each permutation.

See Also

emmixwire scores.wire.

Examples

data(hedenlc)
dat<-hedenlc[seq_len(100),] #for speed
set.seed(123456)

ret <-emmixwire(dat, g=3, ncov=3, nvcov=1, n1=7, n2=8,
   debug=1, itmax=20, epsilon=1e-5)

###calculate the W_j
wj <- scores.wire(ret, contrast=c(1, -1))

### the null distribution of W_j
wj0 <- wj2.permuted(dat, ret, nB=19)
### the p-values of W_j
pv  <- pvalue.wire(wj, wj0)

andrewthomasjones/EMMIXcontrasts documentation built on June 26, 2019, 5:46 a.m.