R devel - Aug 2003 - unexpected behaviour in lm() (PR#3657)

Full_Name: karl-heinz ott
Version: 1.6
OS: wds
Submission from: (NULL) (165.89.84.90)



R : Copyright 2002, The R Development Core Team
Version 1.6.1  (2002-11-01)

R is free software and comes with ABSOLUTELY NO WARRANTY.
You are welcome to redistribute it under certain conditions.
Type `license()' or `licence()' for distribution details.

R is a collaborative project with many contributors.
Type `contributors()' for more information.

Type `demo()' for some demos, `help()' for on-line help, or
`help.start()' for a HTML browser interface to help.
Type `q()' to quit R.
> lm(c(1:10) ~ c(1:10))
Call:
lm(formula = c(1:10) ~ c(1:10))

Coefficients:
(Intercept)  
        5.5  
> lm(c(1:10)+1 ~ c(1:10))
Call:
lm(formula = c(1:10) + 1 ~ c(1:10))

Coefficients:
(Intercept)      c(1:10)  
          1            1  
> lm(c(1:10)+1-1 ~ c(1:10))
Call:
lm(formula = c(1:10) + 1 - 1 ~ c(1:10))

Coefficients:
(Intercept)      c(1:10)  
          0            1  
> lm(c(1:10)+0 ~ c(1:10))
Call:
lm(formula = c(1:10) + 0 ~ c(1:10))

Coefficients:
(Intercept)      c(1:10)  
          0            1  
> lm(c(1:10)+0 ~ c(1:10))
Call:
lm(formula = c(1:10) + 0 ~ c(1:10))

Coefficients:
(Intercept)      c(1:10)  
          0            1  
> lm(numeric(c(1:10)) ~ c(1:10))
Error in model.frame(formula, rownames, variables, varnames, extras, extranames,
 : 
        variable lengths differ
> lm(as.numeric(c(1:10)) ~ c(1:10))
Call:
lm(formula = as.numeric(c(1:10)) ~ c(1:10))

Coefficients:
(Intercept)      c(1:10)  
          0            1  
> lm(c(2:11)) ~ c(1:10))
Error: syntax error
> lm(c(2:11) ~ c(1:10))
Call:
lm(formula = c(2:11) ~ c(1:10))

Coefficients:
(Intercept)      c(1:10)  
          1            1  
> lm(c(2:11) ~ c(1:10))
Call:
lm(formula = c(2:11) ~ c(1:10))

Coefficients:
(Intercept)      c(1:10)  
          1            1  
> plot(c(2:11) ~ c(1:10))
> lm(c(1:10) ~ c(1:10))
Call:
lm(formula = c(1:10) ~ c(1:10))

Coefficients:
(Intercept)  
        5.5  
> lm(c(1:10)+0 ~ c(1:10))
Call:
lm(formula = c(1:10) + 0 ~ c(1:10))

Coefficients:
(Intercept)      c(1:10)  
          0            1  
> lm(c(1:10)+0 ~ c(1:10))
Call:
lm(formula = c(1:10) + 0 ~ c(1:10))

Coefficients:
(Intercept)      c(1:10)  
          0            1  
> lm(as.numeric(c(1:10)) ~ c(1:10))
Call:
lm(formula = as.numeric(c(1:10)) ~ c(1:10))

Coefficients:
(Intercept)      c(1:10)  
          0            1  
> lm(as.numeric(c(1:10)) ~ c(1:10))
Call:
lm(formula = as.numeric(c(1:10)) ~ c(1:10))

Coefficients:
(Intercept)      c(1:10)  
          0            1  
>

ottk@bms.com writes:
> > lm(c(1:10) ~ c(1:10))
> 
> Call:
> lm(formula = c(1:10) ~ c(1:10))
> 
> Coefficients:
> (Intercept)  
>         5.5  
Unexpected: yes, bug: no.

Thing is that the lm logic removes the resonse from the set of
predictors. This is to support stuff like "y ~ .", I believe.

You get the same effect from
> x <- c(1:10)
> lm(x~x)
Call:
lm(formula = x ~ x)

Coefficients:
(Intercept)
        5.5


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

On Mon, 4 Aug 2003 ottk@bms.com wrote:
>
> > lm(c(1:10) ~ c(1:10))
>
> Call:
> lm(formula = c(1:10) ~ c(1:10))
>
> Coefficients:
> (Intercept)
>         5.5
It's a feature. Unfortunately it's not a well documented feature.  A
variable on the right-hand side of the formula is removed if it has the
same name as the left-hand side of the formula.  This is done in
model.matrix.

It's a feature because
 - it's always been done that way
 - it allows  y~. to specify all the columns of a data frame *except* y

It needs to be documented. Unfortunately the logical place is
help(model.matrix), where no-one will look.

	-thomas