Re: [Rd] Bug in poly() (PR#11243)

From: <russell-lenth_at_uiowa.edu>
Date: Fri, 16 May 2008 00:55:18 +0200 (CEST)


I don't understand why this is a bug in usage. Is it because the 2nd argument is not named? I get the same behavior if I do name it:



[R version 2.6.2 (2008-02-08), Windows XP Pro]
R> x = rep(1:4,3)
R> y = (1:12)^1.5
R> lm(y ~ poly(x, degree=10))

Call:
lm(formula = y ~ poly(x, degree = 10))

Coefficients:

            (Intercept)   poly(x, degree = 10)1   poly(x, degree = 10)2
               18.39370                14.21385                 0.58588
  poly(x, degree = 10)3   poly(x, degree = 10)4   poly(x, degree = 10)5
               -0.01770                 3.34767               -11.46388
  poly(x, degree = 10)6   poly(x, degree = 10)7   poly(x, degree = 10)8
                0.51178                 0.44296                12.47199
  poly(x, degree = 10)9  poly(x, degree = 10)10
              -28.38972                18.47439

Is there a case where we *would* want a 10th degree polynomial fitted to only 4 distinct x values? A simple modification [changing 'length(x)' to 'length(unique(x))' in 2 places] seems to fix this:



R> mypoly

    ...

     if (raw) {
         if (degree >= length(unique(x)))
             stop("'degree' must be less than number of points")
    ...
     if (is.null(coefs)) {
         if (degree >= length(unique(x)))
             stop("'degree' must be less than number of points")
    ...

R> lm(y ~ mypoly(x, degree=10))
Error in mypoly(x, degree = 10) :

   'degree' must be less than number of points


Russ

-- 
Russell V. Lenth, Professor
Department of Statistics
   & Actuarial Science            (319)335-0814    FAX (319)335-3017
The University of Iowa           russell-lenth_at_uiowa.edu
Iowa City, IA 52242  USA         http://www.stat.uiowa.edu/~rlenth/

______________________________________________
R-devel_at_r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
Received on Fri 16 May 2008 - 08:34:50 GMT

Archive maintained by Robert King, hosted by the discipline of statistics at the University of Newcastle, Australia.
Archive generated by hypermail 2.2.0, at Fri 16 May 2008 - 19:31:35 GMT.

Mailing list information is available at https://stat.ethz.ch/mailman/listinfo/r-devel. Please read the posting guide before posting to the list.

list of date sections of archive