[R] Newb question re. read.table...

[Rolf Turner]

On 4/03/2010, at 10:50 AM, Steven DeRose wrote:

> I'm trying to get started with R, on Ubuntu. I worked through the
> tutorial, and have made a small tab-delimited subset of my data to try
> out (10 cases with about a dozens variables). But I can't seem to figure
> out how to actually refer to variables. I presume I'm missing something
> very simple, but quite a while searching the doc and the FAQ haven't
> helped me.
> 
> I'm loading the data with
>    con <- read.table("tiny.txt", header=TRUE)
> 
> The first record is a set of variable names, tab-separated like the rest
> of the rows. There are no row labels, thus the same number of
> tab-delimited fields in the header record and the following records. The
> read.table returns silently, and I can get a reasonable summary(con).
> But if I try something like plot(rel,len), where rel and len are two of
> the labels from the header row, I get
> 
>     Error in plot(rel, len) : object 'rel' not found
> 
> I've tried many variations (different variables, adding "con." on the
> front, quoting, using field numbers instead of names, etc. I've also
> read what I can find on read.table, but I'm clearly missing some basic
> thing....
> 
> Can somebody put me back on the right track? Is there some additional
> thing I have to do to make this into a "real" frame, or to bind
> variables names to header names, or something like that?
> 
> Thanks, and sorry for being dense....

You haven't got ``rel'' and ``len'' in your workspace.  They are
columns (components) of the data frame object ``con'' which ***is***
in your workspace.

There are various ways to access components of a data frame:

	* plot(con$rel,con$len)
	* plot(con[["rel"]],con[["len"]])
	* plot(con[,"rel"],con[,"len"])
	* with(con, plot(rel,len))

You should read enough R documentation so that you understand the
under-pinnings of these various syntaxes.

The last one, ``with(con ...'' turns (temporarily) the data frame ``con''
into a data base on your search path.  The components ``rel'' and ``len''
are then objects in this data base and thereby become accessible.

A fifth way to proceed (***NOT*** recommended) is to do:

	attach(con)
	plot(rel,len)
	detach(con)

This is something like the ``with'' solution; with() automates the
attaching and detaching.

The difference is that if you ***have*** got objects ``rel'' and ``len''
in your workspace (and these are different from the columns of ``con'')
then the attach()---detach() procedure will use the objects in your
workspace.  They ``mask'' the columns of ``con''.  The with() procedure
is not beset with this problem.

HTH

	cheers,

		Rolf Turner

######################################################################
Attention:\ This e-mail message is privileged and confid...{{dropped:9}}