R devel - Jun 1998 - R-beta: bugs and problems

A non-text attachment was scrubbed...
Name: not available
Type: text
Size: 1262 bytes
Desc: not available
Url :
https://stat.ethz.ch/pipermail/r-help/attachments/19980602/3cf08662/attachment.pl

Jim Lindsey <jlindsey at luc.ac.be> writes:
> If an MS-DOS file with CRs still in place is (accidently) read into
> R using scan(), an extra zero is added to each line. This can be very
> embarassing!
Are you blaming us or yourself, then? ;)

This looks fairly easy to fix.
> 
> In manipulating repeated measurements data, I generally require
> matrices. Unfortunately, matrix factors create a lot of problems.
> Consider the following simple examples:
This has been changed for 0.62, but I'm afraid not in the direction
you'd want, but in the direction of S-plus compatibility. Basically,
matrix factors don't exist anymore: 
> > x <- as.factor(c("F","N",NA))
> > y <- cbind(x,x)
> > y[,1]
> Error: length of "levels" vector and number of levels differ
> > attributes(y)$levels <- NULL
> > is.factor(y)
> TRUE
> > y[,1]
> F N NA
R> x <- as.factor(c("F","N",NA))
R> y <- cbind(x,x)
R> y[,1]
[1]  1  2 NA
R> is.factor(y)
[1] FALSE
> > rbind(NULL,gl(2,1,2))
> Error in rbind(NULL,gl(2,1,2)): incompatible factors
R> rbind(NULL,gl(2,1,2))
     [,1] [,2]
[1,]    1    2
> 
> 3. To obtain the codes of a factor variable, one uses as.numeric().
> However, this function destroys the properties of the matrix, turning
> it into a vector.
Well...now it also works the other way...:

R> y
     [,1] [,2]
[1,]    1    1
[2,]    2    2
[3,]   NA   NA
R> as.factor(y)
[1] 1  2  NA 1  2  NA
Levels:  1 2 
R> factor(y)
[1] 1  2  NA 1  2  NA
Levels:  1 2 

Actually, a factor can have the dim attribute, but then it doesn't
show up, except in matrix ops and apply where there's an implicit
conversion to integer:

R> r
     [,1] [,2]
[1,]    1    3
[2,]    2    4
R> f<-factor(r)
R> f
[1] 1 2 3 4
Levels:  1 2 3 4 
R> dim(f)<-dim(r)
R> f
[1] 1 2 3 4
Levels:  1 2 3 4 
R> f%*%r
     [,1] [,2]
[1,]    7   15
[2,]   10   22

All this is exactly the same in Splus.

-- 
   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)             FAX: (+45) 35327907

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

>>>>> "Jim" == Jim Lindsey <jlindsey at
luc.ac.be> writes:
    Jim>  .....

    Jim> 3. To obtain the codes of a factor variable, one uses
    Jim> as.numeric().  However, this function destroys the properties of
    Jim> the matrix, turning it into a vector.

as.numeric(.) behaves so consistently (also in S-plus);
it has been a `pain' in other situations, such as passing a numeric matrix
to C or Fortran and wanting the result to still be a matrix.

The ``well-known'' (:-) solution is to use

	storage.mode(x) <- "numeric"

instead, which is rather clumy, especially when passing arguments to
.C(..).
I've seen
	As.numeric <- function(x) { storage.mode(x) <- "numeric"; x }
solving the above problem.

What do you (R-help readers !) think?
Should we add this (and corresponding  As.integer, As.character, ....)
to R ?
Or rather use a different function name?
However, not "as.numeric" itself! 
For such a basic function, we usually don't want to be unnecessarily
incompatible. User functions written for both R and S should get the same
thing for as.numeric(.).

Martin Maechler <maechler at stat.math.ethz.ch>			<><
Seminar fuer Statistik, ETH-Zentrum SOL G1;	Sonneggstr.33
ETH (Federal Inst. Technology)	8092 Zurich	SWITZERLAND
phone: x-41-1-632-3408		fax: ...-1086
http://www.stat.math.ethz.ch/~maechler/

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

On Tue, 2 Jun 1998, Jim Lindsey wrote:
> 
> 3. To obtain the codes of a factor variable, one uses as.numeric().
> However, this function destroys the properties of the matrix, turning
> it into a vector.
>   Factors are very useful, but they are presently a mess for anything
> except simple applications in lm/glm.

You can use codes() to extract the codes of a factor variable while
keeping the matrix structure
R> matrix(factor(1:10),2,5)->a
R> codes(a)
     [,1] [,2] [,3] [,4] [,5]
[1,]    1    3    5    7    9
[2,]    2    4    6    8   10


The other problems need more fixing.  


Thomas Lumley
UW Biostatistics

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._