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.giraud...@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]] ______________________________________________ R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.