[R] glm.nb and Error in x[good, , drop = FALSE] * w : non-conformable arrays

[Patrick Giraudoux]

Many thanks Ivan ! This is fairly clear to me, now... When I dumped the 
data.frame, I found strange to have a "table" declaration for deg, but 
was not able to judge if it was a problem or not (I would have expected 
something as "numeric")
Your workaround is fine to me (I do not need to carry on with table 
attributes).
Best,
Patrick



Le 21/04/2023 à 10:08, Ivan Krylov a écrit :
> On Fri, 21 Apr 2023 09:02:37 +0200
> Patrick Giraudoux<patrick.giraudoux using univ-fcomte.fr>  wrote:
>
>> I meet an error with glm.nb that I cannot explain the origin (and
>> find a fix). The model I want to fit is the following:
>>
>> library(MASS)
>>
>> glm.nb(deg~offset(log(durobs))+zone,data=db)
>>
>> and the data.frame is dumped below.
> Thank you for providing both the code and a small piece of data that
> reproduces the error!
>
> (It almost worked. Your mailer automatically generated a plain text
> version of the e-mail and put Unicode non-breaking spaces in there. R
> considers it a syntax error to encounter any of the various Unicode
> space-like characters outside string literals.)
>
>> deg = structure(c(0, 1, 0, 3, 0, 1, 0, 2, 1, 0, 3, 0, 0, 0, 4, 1, 0,
>> 0, 0, 0, 4, 0, 0, 0, 4, 3, 2, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0,
>> 0, 3, 2, 3, 0, 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 2,
>> 0, 0, 0, 2, 1, 0, 2, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 2,
>> 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 2, 1, 0, 0, 1, 3, 2,
>> 1, 2, 0, 0, 0, 0, 0, 1, 0, 0, 2, 0, 0, 0, 0, 0, 1, 2, 0, 0, 0, 0, 2,
>> 1, 1, 0), dim = 135L, class = "table")
> The problem is that `deg` is a table, which ends up making the
> effective weights a table too. tables are arrays, and element-wise
> product rules are stricter for them than for plain matrices. The code
> makes use of the ability to take an element-wise product between a
> matrix and a vector of the same length as the number of rows in the
> matrix:
>
> matrix(1:12, 4) * 1:4 # works
> matrix(1:12, 4) * as.array(1:4) # results in the same error
>
> # the right way to take products with an array is to make sure that the
> # shapes match exactly
> matrix(1:12, 4) * as.array(cbind(1:4, 1:4, 1:4))
>
> One possible solution is to to remove all attributes from db$deg:
>
> db$deg <- as.vector(db$deg)
>
> This way the values of the expressions involved in glm.fit end up being
> of the expected type, and the function completes successfully.
>

	[[alternative HTML version deleted]]