2
% This file is a component of the package 'sn' for R
3
% copyright (C) 2013 Adelchi Azzalini
4
\title{Package `sn': summary information}
10
\concept{skew-elliptical distribution}
11
\concept{skew-normal distribution}
13
\title{Package \pkg{sn}: overview}
6
This package provides functions related to the skew-normal (SN) and
7
the skew-t (ST) probability distributions, both for the univariate
8
and for the the multivariate case, including regression models.
11
% The functions of the scalar case section are:
12
% \code{dsn}, \code{psn}, \code{qsn}, \code{rsn}, \code{T.Owen},
13
% \code{cp.to.dp}, \code{dp.to.cp}, \code{zeta},
14
% \code{gamma1.to.lambda}, \code{sn.cumulants}, \code{sn.em},
15
% \code{sn.2logL.profile}, \code{sn.mle}, \code{sn.dev},
19
% The functions of the multivariate section are: \code{dmsn},
20
% \code{rmsn}, \code{plot.dsn2}, \code{msn.quantities},
21
% \code{msn.conditional}, \code{msn.marginal}, \code{plot.msn.cond},
22
% \code{msn.fit}, \code{msn.mle}, \code{msn.dev}, \code{msn.dev.grad},
23
% \code{msn.moment.fit}, \code{num.deriv}.
25
The package includes several functions, a number of which are intended
26
as services to other functions, not really to be called directly by the
27
users. The following ones are those more relevant for practical use.
29
\emph{SN distribution:}
30
\code{\link{dsn}}, \code{\link{psn}}, \code{\link{qsn}},
31
\code{\link{rsn}}, \code{\link{cp.to.dp}}, \code{\link{dp.to.cp}},
32
\code{\link{sn.mle}}, \code{\link{sn.em}},
33
%\code{\link{sn.mle.grouped}}
34
for the univariate case, and
35
\code{\link{dmsn}}, \code{\link{pmsn}}, \code{\link{rmsn}},
36
\code{\link{msn.fit}}, \code{\link{msn.mle}}, \code{\link{dsn2.plot}}
37
for the multivariate case.
40
\emph{ST distribution:}
41
\code{\link{dst}}, \code{\link{pst}}, \code{\link{rst}}, \code{\link{st.mle}},
42
%\code{\link{st.mle.grouped}}
43
for the univariate case,
44
and \code{\link{dmst}},\code{\link{pmst}}, \code{\link{rmst}},
45
\code{\link{mst.fit}}, \code{\link{mst.mle}}, \code{\link{dst2.plot}}
46
for the multivariate case.
48
It is suggested that a user starts by reading the documentation of
49
(some of) these functions.
15
The \pkg{sn} package provides facilities to define and manipulate
16
probability distributions of the skew-normal (\acronym{SN}) family and
17
some related ones, notably the skew-\eqn{t} (\acronym{ST}) family,
18
and to apply connected statistical methods for data fitting and
19
diagnostics, in the univariate and the multivariate case.
22
\section{A substantial upgrade}{The first version of the package has been
23
written in 1997 (on CRAN since 1998); subsequent versions have evolved
24
gradually up to version 0.4-18 in May 2013. The present \sQuote{version 1}
25
of the package is a substantial re-writing of the earlier \sQuote{version 0}.
26
Differences between \sQuote{version 0} and \sQuote{version 1} concern the core
27
computational and graphical part as well as the user interface.
28
The S4 protocol for classes and methods has been adopted.
30
Broadly speaking, the available tools can be divided in two groups: the
31
probability section and the statistics section. For a quick start, one
32
could look at their key functions, \code{\link{makeSECdistr}} and
33
\code{\link{selm}}, respectively, and from here explore the rest.
34
In the probability section, one finds also functions \code{\link{dsn}},
35
\code{\link{dst}}, \code{\link{dmsn}} and others alike; these functions
36
existed also in \sQuote{version 0} and their working is still very much
37
the same (not necessarily so their code).
39
The upgrade to \sQuote{version 1} appears more or less at the time when the
40
companion book by Azzalini and Capitanio (2014) is published. Although the two
41
projects are formally separate, they adopt the same notation, terminology
42
and logical frame. This matching and the numerous references in the software
43
documentation to specific sections of the book for background information
44
should facilitate familiarizing with these tools.}
46
\section{Backward Compatibility}{%
48
There is a partial backward compatibility of \sQuote{version 1} versus
49
\sQuote{version 0}. Some functions of the older version would work as before
50
with virtually no change; a wider set arguments is now allowed. Functions
51
\code{\link{dsn}}, \code{\link{dst}}, \code{\link{dmsn}} and alike fall in
52
this category: typically, the names of the arguments have been altered, but
53
they work as before if called with unnamed arguments; similar cases are
54
\code{\link{msn.mle}}, \code{\link{sn.cumulants}} and \code{\link{T.Owen}}.
55
Notice, however, that \code{\link{msn.mle}} and other fitting functions have
56
effectively been subsumed into the more general fitting function
59
A second group of functions will work with little or even minimal changes.
60
Specific examples are functions \code{sn.mle} and \code{sn.mle} which have
61
become \code{\link{sn.mple}} and \code{\link{st.mple}}, with some additional
62
arguments (again, one can achieve the same result via \code{\link{selm}}) and
63
\code{dp.to.cp}, which has been replaced by the more general function
66
Finally, some functions are not there any longer, with no similarly-working
67
functions in the new version. The more prominent set of cases is represented
68
by the functions for computing profile log-likelihoods. There is a long-term
69
plan to re-instate similar facilities, possibly in a more flexible form, but
70
not in the near future. }
51
73
\section{Requirements}{
53
A few functions make use of package \code{mnormt}:
54
\code{psn}, \code{pmsn}, \code{pmst}.
74
\R version 2.15-3 or higher, plus packages \pkg{mnormt},
75
\pkg{numDeriv}, \pkg{stats4} and \pkg{methods} in addition to `standard'
76
packages (\pkg{graphics}, etc.)
57
The version level of the package is given by the command
58
\code{print(.sn.version)}.
80
The command \code{citation("sn")} indicates, among other information,
81
the running version of the package.
59
82
The most recent version of the package can be obtained from
60
83
the WWW page: \url{http://azzalini.stat.unipd.it/SN}
61
which also contains other related material
84
which also contains other related material.
64
Adelchi Azzalini, Dipart. Scienze Statistiche, Universit� di Padova, Italia.
65
Please send comments, error reports, etc. to the author whose WWW page
87
\section{Author}{Adelchi Azzalini.
88
% Dipart. Scienze Statistiche, Università di Padova, Italia.
89
Please send comments, error reports, etc. to the author whose web page
66
90
is \url{http://azzalini.stat.unipd.it/}.
69
This package and its documentation are usable under the terms of the
70
"GNU General Public License", a copy of which is distributed with the
71
package. While the software is freely usable, it would be appreciated
94
This package and its documentation are usable under the terms of
95
the \dQuote{GNU General Public License} version 3 or version 2,
96
as you prefer; a copy of them is available from
97
\url{http://www.R-project.org/Licenses/}.
99
While the software is freely usable, it would be appreciated
72
100
if a reference is inserted in publications or other work
73
101
which makes use of it; for this purpose, see the command
74
102
\code{citation("sn")}.
76
\section{Acknowledgements}{The package has evolved through several
77
versions, developed over some years.
79
For versions up to 0.20, the following people and institutions
80
have contributed. Many thanks go to Antonella Capitanio for testing
81
the procedures, and to Brian Ripley, Kurt Hornik and Martin Maechler
82
for useful advice on R.
83
The function \code{num.deriv2} is based on a similar
84
function written by Monica Chiogna. The first version of this software
85
and part of the associated theoretical work has been developed
86
while the author was at the Nuffield College, Oxford, under the Jemolo
87
Fellowship scheme; the generous support of the college is gratefully
89
Additional support for the development of the theoretical research work has
90
been provided by the "Consiglio Nazionale delle Ricerche" of Italy, grant
93
Versions 0.21 to 0.30 have been supported by "MIUR", Italy, under grant
94
scheme PRIN 2000. Again, thanks to Antonella Capitanio for additional
95
testing of those versions of the package.
106
Azzalini, A. with the collaboration of Capitanio, A. (2014).
107
\emph{The Skew-Normal and Related Families}.
108
Cambridge University Press, IMS Monographs series.
97
111
\keyword{multivariate}
98
112
\keyword{distribution}
99
114
\keyword{regression}