permtest {zebu}

Permutation test for local and global association measures

Description

Permutation test: statistical significance of local and global association measures

Usage

permtest(x, nb = 1000L, group = as.list(colnames(x$data$pp)), p_adjust = "BH")

Arguments

x

lassie S3 object.

nb

number of resampling iterations.

group

list of column names specifying which columns should be permuted together. This is useful for the multivariate case, for example, when there is many dependent variables and one independent variable. By default, permutes all columns separately.

p_adjust

multiple testing correction method. (see p.adjust.methods for a list of methods).

Value

permtest returns an S3 object of class lassie and permtest. Adds the following to the lassie object x:

• global_p: global association p-value.

• local_p: array of local association p-values.

• global_perm: numeric global association values obtained with permutations.

• local_perm: matrix local association values obtained with permutations. Column number correspond to positions in local association array after converting to numeric (e.g. local_perm[, 1] corresponds to local[1]).

• perm_params: parameters used when calling permtest (nb and p_adjust).

See Also

lassie

Examples

# Calling lassie on cars dataset
las <- lassie(cars, continuous = colnames(cars))

# Permutation test using default settings
permtest(las, nb = 30) # keep resampling low for example