R devel - Mar 2003 - question on latticeParseFormula (PR#2602)

This feels like inconsistent behavior.  latticeParseFormula works the
way I anticipated for factor, but not for ordered.  I want the
behavior I see with tmp2, but not with tmp.  My next step is to use
the right.name to isolate the tmp2[,c("a","b")] columns.


tmp <- data.frame(y=(1:12)+.1,
                  a=factor(rep(1:3,4)),
                  b=ordered(rep(1:4, c(3,3,3,3))))
latticeParseFormula(as.formula(y ~ a + b), data=tmp)

tmp2 <- data.frame(y=(1:12)+.1,
                  a=factor(rep(1:3,4)),
                  b=factor(rep(1:4, c(3,3,3,3))))
latticeParseFormula(as.formula(y ~ a + b), data=tmp2)

> version
         _              
platform i386-pc-mingw32
arch     i386           
os       mingw32        
system   i386, mingw32  
status                  
major    1              
minor    6.2            
year     2003           
month    01             
day      10             
language R              
> tmp <- data.frame(y=(1:12)+.1,
+                   a=factor(rep(1:3,4)),
+                   b=ordered(rep(1:4, c(3,3,3,3))))
> latticeParseFormula(as.formula(y ~ a + b), data=tmp)
Error in a + b : non-numeric argument to binary operator
> 
> tmp2 <- data.frame(y=(1:12)+.1,
+                   a=factor(rep(1:3,4)),
+                   b=factor(rep(1:4, c(3,3,3,3))))
> latticeParseFormula(as.formula(y ~ a + b), data=tmp2)
$left
 [1]  1.1  2.1  3.1  4.1  5.1  6.1  7.1  8.1  9.1 10.1 11.1 12.1

$right
 [1] NA NA NA NA NA NA NA NA NA NA NA NA

$condition
NULL

$left.name
[1] "y"

$right.name
[1] "a + b"
> 
Rich

The fact that an attempt at adding an unordered and an ordered factor gives an 
error 

1. is not a bug (at least I can't imagine why it would be so)
2. has nothing to do with latticeParseFormula

Also, I very much doubt that getting a bunch of NA's and a warning (which
you
seem to have omitted from your output) is really what you anticipated when 
you added two (unordered) factors.

Please first report what you think are bugs on r-devel before filing bug 
reports.


On Monday 03 March 2003 06:04 pm, rmh@surfer.sbm.temple.edu
wrote:
> This feels like inconsistent behavior.  latticeParseFormula works the
> way I anticipated for factor, but not for ordered.  I want the
> behavior I see with tmp2, but not with tmp.  My next step is to use
> the right.name to isolate the tmp2[,c("a","b")]
columns.
>
>
> tmp <- data.frame(y=(1:12)+.1,
>                   a=factor(rep(1:3,4)),
>                   b=ordered(rep(1:4, c(3,3,3,3))))
> latticeParseFormula(as.formula(y ~ a + b), data=tmp)
>
> tmp2 <- data.frame(y=(1:12)+.1,
>                   a=factor(rep(1:3,4)),
>                   b=factor(rep(1:4, c(3,3,3,3))))
> latticeParseFormula(as.formula(y ~ a + b), data=tmp2)
>
> > version
>
>          _
> platform i386-pc-mingw32
> arch     i386
> os       mingw32
> system   i386, mingw32
> status
> major    1
> minor    6.2
> year     2003
> month    01
> day      10
> language R
>
> > tmp <- data.frame(y=(1:12)+.1,
>
> +                   a=factor(rep(1:3,4)),
> +                   b=ordered(rep(1:4, c(3,3,3,3))))
>
> > latticeParseFormula(as.formula(y ~ a + b), data=tmp)
>
> Error in a + b : non-numeric argument to binary operator
>
> > tmp2 <- data.frame(y=(1:12)+.1,
>
> +                   a=factor(rep(1:3,4)),
> +                   b=factor(rep(1:4, c(3,3,3,3))))
>
> > latticeParseFormula(as.formula(y ~ a + b), data=tmp2)
>
> $left
>  [1]  1.1  2.1  3.1  4.1  5.1  6.1  7.1  8.1  9.1 10.1 11.1 12.1
>
> $right
>  [1] NA NA NA NA NA NA NA NA NA NA NA NA
>
> $condition
> NULL
>
> $left.name
> [1] "y"
>
> $right.name
> [1] "a + b"
>
>
>
> Rich
>
> ______________________________________________
> R-devel@stat.math.ethz.ch mailing list
> http://www.stat.math.ethz.ch/mailman/listinfo/r-devel

I didn't get a warning.  I got exactly what I sent.

The NA doesn't bother me since I am interested in parsing the expression,
specifically to see what is to the left of the "~" and what is to the
right of the "~".   I am interested in components left.name and
right.name,
.I am not interested in left and right.  My inference from your response
is that I should work with the parse tree from 
  parse(text="y ~ a + b")
rather than continue pushing latticeParseFormula.

On the other issue that you raised, I am surprised to learn that R
can add factors to factors and ordered to ordered.  I didn't expect either
case.  However, given that it can do something, I still would expect
that it could add them to each other by recognizing that ordered inherits
from factor.
> ordered(1:3) + ordered(3:5)
[1] NA NA NA
> factor(1:3) + factor(3:5)
[1] NA NA NA
> factor(1:3) + ordered(3:5)
Error in factor(1:3) + ordered(3:5) : non-numeric argument to binary
operator
>
thank you for the prompt response.
Rich

I looked again.  I had turned off the warning messages.  I intended
to have them off only inside the loop and turn them on again afterwards.
I will fix that.

Rich