R packages - Jan 2019 - permuco: permutation tests and multiple comparisons.

Hello everyone,

We present you the permuco package, which has 2 main purposes: PERmutation tests
and MUltiple COmparisons.

First, the package has functions for permutation tests for parameters in linear
models with nuisance variables. Several permutation methods exist in the
literature to reduce the effect of nuisance variables and the permuco package
allows to use them for t statistics in regression model, F tests in type 3
ANOVA, and F tests in repeated measures ANOVA. The 2 functions aovperm() and
lmperm() perform these tests and their usage is similar to the parametric
counterpart aov() and lm().

Secondly, it uses multiple comparisons procedures with permutation tests to
compare signals time by time. This problem is common when analysing
electroencephalography (or fMRI) data where the response variables are EEG
signals recorded for each experimental setting. In these experiments, the number
of tests is equal to the number of time points of the signal which is typically
in the thousands. The permuco package implements the state-of-the-art multiple
comparisons procedures which uses permutations, like the cluster-mass test or
the threshold-free cluster-enhancement. These procedures are powerful when the
effects appear in clusters and they will control the family wise error rate
(FWER) under the null hypothesis.

All these procedures are explained and exemplified in detail in a vignette.

We are waiting for your feedback in order to improve the next releases of
permuco.

Olivier Renaud & Jaromil Frossard

University of Geneva

URL:
https://CRAN.R-project.org/package=permuco
Maintainer:
jaromil.frossard at unige.ch<mailto:jaromil.frossard at unige.ch>

<mailto:jaromil.frossard at unige.ch>


	[[alternative HTML version deleted]]