[Rd] weights in glm (PR#8720)
ripley at stats.ox.ac.uk
ripley at stats.ox.ac.uk
Tue Mar 28 19:39:32 CEST 2006
First, R-bugs is not for asking questions, only for reporting things you
are *certain* are bugs: see the R FAQ.
Why are you using weights to omit cases? If you had used subset, this
would have worked. The problem is the use of zero weights, which are not
intended to be used in this way. We can fix this up, but it is not the
correct way to use glm.
On Tue, 28 Mar 2006, robert.pusz at wp.pl wrote:
> Full_Name: Robert Pusz
> Version: 2.2.1
> OS: Windows
> Submission from: (NULL) (157.25.9.126)
>
>
> Hello,
> In my opinion something is wrong with 'weights' option in glm.
> My code is following:
> ###begin of the code####
> cl <- c(5012, 106, 3410, 5655, 1092, 1513, 557, 1351, 3133, 2063, 3257,
> 4179, 5582, 5900, 8473, 4932, 3463, 5596, 2262, 0, 2638, 1111,
> 4881, 4211, 6271, 5257, 6926, 6165, 0, 0, 898, 5270, 2268, 5500,
> 6333, 1233, 1368, 0, 0, 0, 1734, 3116, 2594, 2159, 3786, 2917,
> 0, 0, 0, 0, 2642, 1817, 3479, 2658, 225, 0, 0, 0, 0, 0, 1828,
> 0, 649, 984, 0, 0, 0, 0, 0, 0, 599, 673, 603, 0, 0, 0, 0, 0,
> 0, 0, 54, 535, 0, 0, 0, 0, 0, 0, 0, 0, 172, 0, 0, 0, 0, 0, 0,
> 0, 0, 0)
>
> w <- c(1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0,
> 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1,
> 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 1, 0,
> 1, 1, 0, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0,
> 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0)
>
> row <- c(1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10,
> 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10,
> 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10,
> 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10,
> 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10)
>
> col <- c(1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
> 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 5,
> 5, 5, 5, 5, 5, 5, 5, 5, 5, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 7, 7,
> 7, 7, 7, 7, 7, 7, 7, 7, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 9, 9, 9,
> 9, 9, 9, 9, 9, 9, 9, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10)
> Row <- ordered(as.factor(row))
> Col <- ordered(as.factor(col))
> fit <- glm(cl ~ Row + Col, family = quasipoisson, weights = w)
> ###end of the code####
>
> When I have written summary(fit) I have got:
> Estimate Std. Error t value Pr(>|t|)
> (Intercept) 7.64459 NA NA NA
> Row2 -0.05467 NA NA NA
> Row3 0.24541 NA NA NA
> Row4 0.42035 NA NA NA
> Row5 0.43961 NA NA NA
> Row6 0.04532 NA NA NA
> Row7 -0.04881 NA NA NA
> Row8 0.25370 NA NA NA
> Row9 -0.14976 NA NA NA
> Row10 -0.01267 NA NA NA
> Col2 0.69283 NA NA NA
> Col3 0.62603 NA NA NA
> Col4 0.27695 NA NA NA
> Col5 0.06056 NA NA NA
> Col6 -0.19582 NA NA NA
> Col7 -0.83044 NA NA NA
> Col8 -1.27914 NA NA NA
> Col9 -1.93235 NA NA NA
> Col10 -2.49709 NA NA NA
>
> When I omited 'weights=w' above table was filled in with numbers, but the
> results were wrong (because of taking zeros in regression).
>
> Could you tell me what's wrong?
> Kind regards,
> Robert
>
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel
>
>
--
Brian D. Ripley, ripley at stats.ox.ac.uk
Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/
University of Oxford, Tel: +44 1865 272861 (self)
1 South Parks Road, +44 1865 272866 (PA)
Oxford OX1 3TG, UK Fax: +44 1865 272595
More information about the R-devel
mailing list