https://github.com/cran/sn
Tip revision: bc33612e6cc33fcf28f50655cab5f1931985ccde authored by Adelchi Azzalini on 04 April 2023, 17:10:02 UTC
version 2.1.1
version 2.1.1
Tip revision: bc33612
fournum.Rd
% file sn/man/fournum.Rd
% This file is a component of the package 'sn' for R
% copyright (C) 2013-2020 Adelchi Azzalini
%---------------------
\name{fournum}
\encoding{UTF-8}
\alias{fournum}
\concept{quantile}
\concept{median}
\concept{variability}
\concept{asymmetry}
\concept{skewness}
\concept{kurtosis}
\title{Four-number summary of a numeric vector}
\description{
Returns a quantile-based four-number summary of the input data
}
\usage{
fournum(x, na.rm = TRUE, ...)
}
\arguments{
\item{x}{a numeric vector, maybe including \code{NA}s and \code{+/-Inf}'s.
At least 8 not-\code{NA} values are required.
It works with objects which can be coerced to vector.}
\item{na.rm}{ logical;
if \code{TRUE}, all \code{NA} and \code{NaN}s are dropped,
before the statistics are computed.}
\item{\dots}{optional arguments passed to \code{\link[stats]{quantile}}}
}
\details{Function \code{quantile} is used to compute 7 octiles of \code{x},
that is, quantiles of level \code{(1:7)/8}, denoted \code{oct[1:7]},
and derive four summary quantities:
\enumerate{
\item the median, which corresponds to \code{oct[4]},
\item the \sQuote{(coefficient of) quartile deviation}
or semi-interquantile range: \code{(oct[6] - oct[2])/2};
\item the Galton-Bowley measure of asymmetry, that is, skewness:
\code{(oct[6] - 2 * oct[4] + oct[2])/(oct[6] - oct[2])};
\item the Moors measure of kurtosis:
\code{(oct[7] - oct[5] + oct[3] - oct[1])/(oct[6] - oct[2])}
}
The term \sQuote{coefficient of quartile deviation} is adopted from the
Encyclopedia of Statistical Sciences; see the references below.
What is called Galton-Bowley measure here is often named
\sQuote{Bowley's measure}, but some sources attribute it to Francis Galton.
For the Moors measure, see Moors (1988).
}
\value{a vector of length four containing the median, the quartile deviation,
the Galton-Bowley measure and the Moors measure. However,
if \code{x} does not contain at least 8 values (after removing \code{NA}s),
\code{rep(NA,4)} is returned.}
\references{
\sQuote{Quartile deviation, coefficient of}, in:
\emph{Encyclopedia of Statistical Sciences}, 2nd edition (2006).
Editors: Samuel Kotz (Editor-in-Chief), Campbell B. Read, N. Balakrishnan,
Brani Vidakovic. Volume 10, p.6743.
\sQuote{Skewness, Bowleys's measures of}, in:
\emph{Encyclopedia of Statistical Sciences}, 2nd edition (2006).
Editors: Samuel Kotz (Editor-in-Chief), Campbell B. Read, N. Balakrishnan,
Brani Vidakovic. Volume 12, p.7771-7773.
Moors, J.J.A. (1988).
A quantile alternative for kurtosis.
\emph{Source: Journal of the Royal Statistical Society.
Series D (The Statistician)}, Vol. 37, pp. 25-32
}
\author{Adelchi Azzalini}
\note{
Computation of octiles makes real sense only if \code{length(x)} is
substantially larger than 8. }
\seealso{\code{\link[stats]{quantile}}, \code{\link[stats]{fivenum}},
\code{\link[stats]{IQR}} }
\examples{
fournum(datasets::rivers)
}
% Add one or more standard keywords, see file 'KEYWORDS' in the
% R documentation directory.
\keyword{univar}
\keyword{nonparametric}
\keyword{robust}