[R] Problems with rlm

[Liaw, Andy]

You have not given us nearly enough info.  On my WinXP laptop with 512MB
RAM, R-1.8.1, I get:

> system.time(out <- rlm(h ~ c))
[1] 0.03 0.00 0.04   NA   NA
> out
Call:
rlm(formula = h ~ c)
Converged in 8 iterations

Coefficients:
(Intercept)           c 
 -1.4720930  -0.1843265 

Degrees of freedom: 27 total; 25 residual
Scale estimate: 0.541 
> h
         1          2          3          6          7          9         10

-2.7916667 -1.0246838 -2.0681170 -1.3256766 -3.1122708 -2.1444948 -2.5806538

        11         12         13         14         15         16         18

-0.6791551 -0.1278583 -3.7737900 -3.3077561 -2.4615233 -3.4016225 -2.2484673

        21         22         23         25         30         31         35

-0.8938528 -2.2521325 -2.3461999 -2.1405470 -2.7657994 -3.3588822 -2.8144453

        39         40         44         45         47         50 
-3.8521439 -2.1882352 -1.7531826 -3.6464482 -2.2158240 -2.7162693 
> c
        1         2         3         6         7         9        10
11 
 6.000000  2.500000  3.250000  4.000000  2.333333  2.400000  2.750000
2.000000 
       12        13        14        15        16        18        21
22 
 4.000000  5.000000  4.000000  7.000000  4.500000  3.500000  4.000000
5.000000 
       23        25        30        31        35        39        40
44 
 6.000000  3.000000  9.000000 10.000000  8.000000 10.000000  4.000000
6.000000 
       45        47        50 
 4.500000  5.500000  5.000000 

What OS / R version / MASS version (you are using MASS, no?) are you using?
What's your hardware?  Do you have lots of other stuff in the workspace when
you tried to run this?  What does gc() tell you?

Andy

> From: Ed Hagen
> 
> Dear all,
> 
> When calling rlm with the following data, I get an error.  (R 
> v.1.8.1, 
> WinXP Pro 2002 with service pack 1.)
> 
> > d <- na.omit(data.frame(CPRATIO, HEIGHTZ, FAMILYID))
> > c <- tapply(d$CPRATIO, d$FAMILYID, mean)
> > h <- tapply(d$HEIGHTZ, d$FAMILYID, mean)
> > c
>          1         2         3         6         7         9        10
>        11
>   6.000000  2.500000  3.250000  4.000000  2.333333  2.400000  2.750000
> 2.000000
>         12        13        14        15        16        18        21
>        22
>   4.000000  5.000000  4.000000  7.000000  4.500000  3.500000  4.000000
> 5.000000
>         23        25        30        31        35        39        40
>        44
>   6.000000  3.000000  9.000000 10.000000  8.000000 10.000000  4.000000
> 6.000000
>         45        47        50
>   4.500000  5.500000  5.000000
> > h
>           1          2          3          6          7          9
>     10
> -2.7916667 -1.0246838 -2.0681170 -1.3256766 -3.1122708 -2.1444948
> -2.5806538
>          11         12         13         14         15         16
>     18
> -0.6791551 -0.1278583 -3.7737900 -3.3077561 -2.4615233 -3.4016225
> -2.2484673
>          21         22         23         25         30         31
>     35
> -0.8938528 -2.2521325 -2.3461999 -2.1405470 -2.7657994 -3.3588822
> -2.8144453
>          39         40         44         45         47         50
> -3.8521439 -2.1882352 -1.7531826 -3.6464482 -2.2158240 -2.7162693
> > out <- rlm(h ~ c)
> Error in model.matrix.default(mt, mf, contrasts) :
>          cannot allocate vector of length 1077237505
> 
> 
> If I make the same rlm call a second or third time, R usually crashes.
> 
> Any help or suggestions would be greatly appreciated.  These commands
> were run after a much larger script had seemingly run 
> successfully. But,
> I had no problem calling, e.g., rlm(HEIGHTZ ~ CPRATIO), after the same
> script had run.  Although I have made some minor changes in 
> the script, 
> I also didn't have this problem when using R 1.7.
> 
> Many thanks in advance,
> 
> Ed.
> 
> 
> -- 
> Edward H. Hagen                  Institute for Theoretical Biology
> phone: +49/30 2093-8649             Humboldt-Universität zu Berlin
> fax:   +49/30 2093-8801                         Invalidenstraße 43
> http://itb.biologie.hu-berlin.de/~hagen      10115 Berlin, Germany
> 
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://www.stat.math.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide! 
> http://www.R-project.org/posting-guide.html
> 
>