The R Journal Volume 4/1, June 2012

R Foundation For Statistical Computing

Informations

• Publié par: mtoledan
• Publié le: 26 juillet 2012
• Nombre de lectures: 652
• Langue: English
• Poids de l'ouvrage: 4 Mo

Extrait

The Journal
Volume 4/1, June 2012
A peer-reviewed, open-access publication of the R Foundation
for Statistical Computing
Contents
Editorial . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Contributed Research Articles
Analysing Seasonal Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
MARSS: Multivariate Autoregressive State-space Models for Analyzing Time-series Data . . 11
openair – Data Analysis Tools for the Air Quality Community . . . . . . . . . . . . . . . . . . 20
Foreign Library Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
Vdgraph: A Package for Creating Variance Dispersion Graphs . . . . . . . . . . . . . . . . . . 41
xgrid and R: Parallel Distributed Processing Using Heterogeneous Groups of Apple Com-
puters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
maxent: An R Package for Low-memory Multinomial Logistic Regression with Support for
Semi-automated Text Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Sumo: An Authenticating Web Application with an Embedded R Session . . . . . . . . . . . 60
From the Core
Who Did What? The Roles of R Package Authors and How to Refer to Them . . . . . . . . . 64
News and Notes
Changes in R . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 on CRAN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
R Foundation News . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96

2
The Journal is a peer-reviewed publication of the R Foun-
dation for Statistical Computing. Communications regarding this pub-
lication should be addressed to the editors. All articles are licensed un-
der the Creative Commons Attribution 3.0 Unported license (CC BY 3.0,
http://creativecommons.org/licenses/by/3.0/).
Prospective authors will find detailed and up-to-date submission in-
structions on the Journal’s homepage.
Editor-in-Chief:
Martyn Plummer
Editorial Board:
Heather Turner, Hadley Wickham, and Deepayan Sarkar
Editor Programmer’s Niche:
Bill Venables
Editor Help Desk:
Uwe Ligges
Editor Book Reviews:
G. Jay Kerns
Department of Mathematics and Statistics
Youngstown State University
Youngstown, Ohio 44555-0002
USA
[e-mail]
R Journal Homepage:
http://journal.r-project.org/
Email of editors and editorial board:
[e-mail]
The R Journal is indexed/abstracted by EBSCO, DOAJ.
The R Journal Vol. 4/1, June 2012 ISSN 2073-4859

3
Editorial
by Martyn Plummer Interface for R to call arbitrary native functions with-
out the need for C wrapper code. There is also an ar-
Earlier this week I was at the “Premières Rencontres ticle from our occasional series “From the Core”, de-
R” in Bordeaux, the first francophone meeting of R signed to highlight ideas and methods in the words
users (although I confess that I gave my talk in En- of R development core team members. This article
glish). This was a very enjoyable experience, and by Kurt Hornik, Duncan Murdoch and Achim Zeilies
not just because it coincided with the Bordeaux wine explains the new facilities for handling bibliographic
festival. The breadth and quality of the talks were information introduced in R 2.12.0 and R 2.14.0. I
both comparable with the International UseR! con- strongly encourage package authors to read this ar-
ferences. It was another demonstration of the extent ticle and implement the changes to their packages.
to which R is used in diverse disciplines. Doing so will greatly facilitate the bibliometric anal-
As R continues to expand into new areas, we yses and investigations of the social structure of the
see the development of packages designed to ad- R community.
dress the specific needs of different user communi- Elsewhere in this issue, we have articles describ-
ties. This issue of The R Journal contains articles on ing the season package for displaying and analyz-
two such packages. Karl Ropkins and David Carslaw ing seasonal data, which is a companion to the book
describe how the design of the openair package was Analysing Seasonal Data by Adrian Barnett and An-
influenced by the need to make an accessible set of nette Dobson; the Vdgraph package for drawing
tools available for people in the air quality commu- variance dispersion graphs; and the maxent pack-
nity who are not experts in R. Elizabeth Holmes and age which allows fast multinomial logistic regression
colleagues introduce the MARSS package for mul- with a low memory footprint, and is designed for ap-
tivariate autoregressive state-space models. Such plications in text classification.
models are used in many fields, but the MARSS
The R Journal continues to be the journal ofpackage was motivated by the particular needs of re-
record for the R project. The usual end matter sum-searchers in the natural and environmental sciences,
marizes recent changes to R itself and on CRAN. ItThere are two articles on the use of R outside the
is worth spending some time browsing these sec-desktop computing environment. Sarah Anoke and
tions in order to catch up on changes you may havecolleagues describe the use of the Apple Xgrid sys-
missed, such as the new CRAN Task View on dif-tem to do distributed computing on Mac OS X com-
ferential equations. Peter Dalgaard highlighted theputers, and Timothy Bergsma and Michael Smith
importance of this area in his editorial for volume 2,demonstrate the Sumo web application, which in-
issue 2.cludes an embedded R session.
Two articles are of particular interest to develop- On behalf of the editorial board I hope you enjoy
ers. Daniel Adler demonstrates a Foreign Function this issue.
The R Journal Vol. 4/1, June 2012 ISSN 2073-4859

4
The R Journal Vol. 4/1, June 2012 ISSN 2073-4859

CONTRIBUTED RESEARCH ARTICLES 5
Analysing Seasonal Data
by Adrian G Barnett, Peter Baker and Annette J Dobson but often very effective analyses, as we describe be-
low.
Abstract Many common diseases, such as the flu More complex seasonal analyses examine non-
and cardiovascular disease, increase markedly stationary seasonal patterns that change over time.
in winter and dip in summer. These seasonal Changing in health are currently
patterns have been part of life for millennia and of great interest as global warming is predicted to
were first noted in ancient Greece by both Hip- make seasonal changes in the weather more extreme.
pocrates and Herodotus. Recent interest has fo- Hence there is a need for statistical tools that can es-
cused on climate change, and the concern that timate whether a seasonal pattern has become more
seasons will become more extreme with harsher extreme over time or whether its phase has changed.
winter and summer weather. We describe a set Ours is also the first R package that includes the
of R functions designed to model seasonal pat- case-crossover, a useful method for controlling for
terns in disease. We illustrate some simple de- seasonality.
scriptive and graphical methods, a more com- This paper illustrates just some of the functions of
plex method that is able to model non-stationary the season package. We show some descriptive func-
patterns, and the case-crossover to control for tions that give simple means or plots, and functions
seasonal confounding. whose goal is inference based on generalised linear
models. The package was written as a companion to
a book on seasonal analysis by Barnett and DobsonIn this paper we illustrate some of the functions
(2010), which contains further details on the statisti-of the season package (Barnett et al., 2012), which
cal methods and R code.contains a range of functions for analysing seasonal
health data. We were motivated by the great inter-
est in seasonality found in the health literature, and
Analysing monthly seasonal pat-the relatively small number of seasonal tools in R (or
other software packages). The existing seasonal tools terns
in R are:
Seasonal time series are often based on data collected
• the baysea function of the timsac package and every month. An example that we use here is the
the decompose and stl functions of the stats monthly number of cardiovascular disease deaths in
package for decomposing a time series into a people aged 75 years in Los Angeles for the years
trend and season; 1987–2000 (Samet et al., 2000). Before we examine
or plot the monthly death rates we need to make
• the dynlm function of the dynlm package and
them more comparable by adjusting them to a com-
thessm of the sspir package for fitting
mon month length (Barnett and Dobson, 2010, Sec-
dynamic linear models with optional seasonal
tion 2.2.1). Otherwise January (with 31 days) will
components;
likely have more deaths than February (with 28 or
29).• thearima function of the stats package and the
In the example below the monthmean functionArima function of the forecast for fit-
is used to create the variable mmean which is theting seasonal components as part of an autore-
monthly average rate of cardiovascular diseasegressive integrated moving average (ARIMA)
deaths standardised to a month length of 30 days. Asmodel; and
the data set contains the population siz