no doubt a dumb question, but..

Yes, I am a newbie at R, but it is not the complex commands in R that 
have me baffled, but simple data commands.  For example, why does 
something like:

 > plot(Girth ~ Height)

*not* work after a command that allegedly loads the data:

 > data(trees)

with the error message:

Error in eval(expr, envir, enclos) : Object "Girth" not found

but does work after the command:

 > attach(trees)

?

As per ?attach:

"By attaching a data frame to the search path it is possible to refer to
the variables in the data frame by their names alone, rather than as
components of the data frame (eg in the example below, height rather
than women$height)."

data(trees) simply loads the dataset, but does not place it in the
search path, which is what attach(trees) does.

Prior to using attach(trees), you would need to use:

plot(trees$Girth ~ trees$Height)

to tell R that the variables Girth and Height are in the data frame
called 'trees'.

HTH,

Marc Schwartz

"Marc" == Marc Schwartz <MSchwartz at MedAnalytics.com>
    on Mon, 08 Nov 2004 22:27:47 -0600 writes:

>> Yes, I am a newbie at R, but it is not the complex
    >> commands in R that have me baffled, but simple data
    >> commands.  For example, why does something like:
    >> 
    >> > plot(Girth ~ Height)
    >> 
    >> *not* work after a command that allegedly loads the data:
    >> 
    >> > data(trees)
    >> 
    >> with the error message:
    >> 
    >> Error in eval(expr, envir, enclos) : Object "Girth" not
    >> found
    >> 
    >> but does work after the command:
    >> 
    >> > attach(trees)
    >> 
    >> ?

    Marc> As per ?attach:

    Marc> "By attaching a data frame to the search path it is
    Marc> possible to refer to the variables in the data frame
    Marc> by their names alone, rather than as components of the
    Marc> data frame (eg in the example below, height rather
    Marc> than women$height)."

    Marc> data(trees) simply loads the dataset, but does not
    Marc> place it in the search path, which is what
    Marc> attach(trees) does.

    Marc> Prior to using attach(trees), you would need to use:

    Marc> plot(trees$Girth ~ trees$Height)

    Marc> to tell R that the variables Girth and Height are in
    Marc> the data frame called 'trees'.

Yes, thank you, Marc!

For the specific case at hand, however, the recommdend way is to use

  plot(Girth ~ Height, data = trees)


Further note that many of us try to avoid attach()ing data frames
(most of the time; not always) and we have provided the nice alternative
    with( <data> ,  <expression_body> ) 

So, for the current example, you could also say

  with(trees, plot(Girth ~ Height))

The main advantage of with(): Only inside it, the components of
'trees' are visible - no need to remember to  detach() ;
see also  help(with) and its examples.

Martin Maechler

<SNIP>

Further note that many of us try to avoid attach()ing data frames
(most of the time; not always) and we have provided the nice alternative
    with( <data> ,  <expression_body> ) 

So, for the current example, you could also say

  with(trees, plot(Girth ~ Height))

The main advantage of with(): Only inside it, the components of
'trees' are visible - no need to remember to  detach() ;
see also  help(with) and its examples.

Right. Thanks Martin. After sending my reply, I kicked myself a few
times for not mentioning with()...especially because it also enables a
standardized mechanism for accessing data frame variables across
functions (not all of which have a 'data = ' argument).

Also, there is a substantial savings with respect to overhead by not
attaching and detaching (manipulating the search path), which I had
noted and benchmarked previously in a post, earlier this year I believe,
that I cannot locate at the moment. 

This can become very important if one is looping (in some fashion) over
data frames or subsets of data frames and therefore avoids the repeated
calls to attach() and detach().

Best regards,

Marc

Yes, I am a newbie at R, but it is not the complex commands in R that
have me baffled, but simple data commands.  For example, why does

something like:

 > plot(Girth ~ Height)

*not* work after a command that allegedly loads the data:

 > data(trees)

with the error message:

Error in eval(expr, envir, enclos) : Object "Girth" not found

but does work after the command:

 > attach(trees)

I have wondered this myself, but one reason may be that more than one data 
table can be loaded at a time.  Quite often data tables will use the same 
names for fields.  It would inconvenient for instance, if you had two tables 
loaded, say one for trees and one for melons with a field named "Girth", and 
tried a plot(Girth ~ Height) command.  So data() more or less accesses that 
table, and attach explicit identifies the table of interest.

JWDougherty