In an Rweave tutorial written for possibly naive R users, I felt it
necessary to explain why
table() had to be used inside with(), whereas other tools like xtabs()
had a data= argument.
with() is quite nice for such cases, but it seems an unnecessary thing
to learn right off.
Before I turn this question into a request for R-devel, is there any
inherent reason why it might be
hard to add a data= argument to table()? I've looked at the code, but
am not enlightened
on this question.
\emph{Example}: Convert the \code{Arthritis} data in case form to a
3-way table of
\code{Treatment} $\times$ \code{Sex} $\times$ \code{Improved}.%
\footnote{
Unfortunately, \codefun{table} does not allow a \code{data} argument to
provide
an environment in which the table variables are to be found. In the
examples in \secref{sec:table} I used \code{attach(mydata)} for this
purpose,
but \codefun{attach} leaves the variables in the global environment,
while \codefun{with} just evaluates the \codefun{table} expression in a
temporary environment of the data.
}
<<convert-ex2,results=verbatim>>Art.tab <-with(Arthritis,
table(Treatment, Sex, Improved))
str(Art.tab)
ftable(Art.tab)
@
--
Michael Friendly Email: friendly AT yorku DOT ca
Professor, Psychology Dept.
York University Voice: 416 736-5115 x66249 Fax: 416 736-5814
4700 Keele Street http://www.math.yorku.ca/SCS/friendly.html
Toronto, ONT M3J 1P3 CANADA
on 02/20/2009 09:11 AM Michael Friendly wrote:> In an Rweave tutorial written for possibly naive R users, I felt it > necessary to explain why > table() had to be used inside with(), whereas other tools like xtabs() > had a data= argument. > > with() is quite nice for such cases, but it seems an unnecessary thing > to learn right off. > Before I turn this question into a request for R-devel, is there any > inherent reason why it might be > hard to add a data= argument to table()? I've looked at the code, but > am not enlightened > on this question. > > \emph{Example}: Convert the \code{Arthritis} data in case form to a > 3-way table of > \code{Treatment} $\times$ \code{Sex} $\times$ \code{Improved}.% > \footnote{ > Unfortunately, \codefun{table} does not allow a \code{data} argument to > provide > an environment in which the table variables are to be found. In the > examples in \secref{sec:table} I used \code{attach(mydata)} for this > purpose, > but \codefun{attach} leaves the variables in the global environment, > while \codefun{with} just evaluates the \codefun{table} expression in a > temporary environment of the data. > } > <<convert-ex2,results=verbatim>>> Art.tab <-with(Arthritis, table(Treatment, Sex, Improved)) > str(Art.tab) > ftable(Art.tab) > @xtabs and other functions (such as modeling and plot functions) that include a 'data' argument have a formula based argument as the means by which you indicate the columns of the data frame to be included. These are then passed to model.frame() internally to create the data frame to be used subsequently by the function. Since model.frame() is evaluated within the environment of the data frame indicated by the 'data' argument if present, it is needed there. It's always dangerous to say always, but in my experience, functions that have a 'data' argument fit the above profile. If table() had a formula method, then having a 'data' argument would make sense. HTH, Marc Schwartz
On Fri, 20 Feb 2009, Michael Friendly wrote:> In an Rweave tutorial written for possibly naive R users, I felt it necessary > to explain why table() had to be used inside with(), whereas other tools > like xtabs() had a data= argument.I think the simple rule is: If there is a "formula", there is (typically) also a "data" argument, otherwise not. There are probably exceptions but So the question should probably be: Why is there xtabs() with interface xtabs(~ x + y, data = mydata) instead of a "formula" method for table() that would allow table(~ x + y, data = mydata) Well, table() is not generic, to some extent probably due to historical reasons. But maybe it would be worth considering a change to this? Z> with() is quite nice for such cases, but it seems an unnecessary thing to > learn right off. > Before I turn this question into a request for R-devel, is there any inherent > reason why it might be > hard to add a data= argument to table()? I've looked at the code, but am not > enlightened > on this question. > > \emph{Example}: Convert the \code{Arthritis} data in case form to a 3-way > table of > \code{Treatment} $\times$ \code{Sex} $\times$ \code{Improved}.% > \footnote{ > Unfortunately, \codefun{table} does not allow a \code{data} argument to > provide > an environment in which the table variables are to be found. In the > examples in \secref{sec:table} I used \code{attach(mydata)} for this purpose, > but \codefun{attach} leaves the variables in the global environment, > while \codefun{with} just evaluates the \codefun{table} expression in a > temporary environment of the data. > } > <<convert-ex2,results=verbatim>>> Art.tab <-with(Arthritis, table(Treatment, Sex, Improved)) > str(Art.tab) > ftable(Art.tab) > @ > > > -- > Michael Friendly Email: friendly AT yorku DOT ca Professor, Psychology > Dept. > York University Voice: 416 736-5115 x66249 Fax: 416 736-5814 > 4700 Keele Street http://www.math.yorku.ca/SCS/friendly.html > Toronto, ONT M3J 1P3 CANADA > > ______________________________________________ > R-help at r-project.org mailing list > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide http://www.R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code. > >