% Generated by roxygen2: do not edit by hand
% Please edit documentation in R/MixMC.R
\name{normalise.logit.empirical}
\alias{normalise.logit.empirical}
\title{Normalise using the logit function in an empirical manner}
\usage{
normalise.logit.empirical(input)
}
\arguments{
\item{input}{Data frame of input compositional data to normalise. Input data
should be proportions 0-1.}
}
\value{
Data normalised using empirical logit. Proportions below 0.5 will be
negative, but output will not tend to infinity for zero or 1 input.
}
\description{
Apply the empirical logit normalisation to a data frame of omics input data.
This is intended to convert compositional data, e.g. proportional data in the
range 0..1, to Euclidean space which is most appropriate for the linear
models. The empirical logit function differs for standard logit normalisation
in that an epsilon factor is added to ensure that function does not tend to
+/- infinity for input values close to 100\% and 0\% respectively. The logit
or empirical logit function will be a more appropriate choice than centred
log-ratio (CLR) for non-OTU data.
}
\examples{
\dontrun{
normalise.logit.empirical(data.proportional)
}
}