[R-sig-ME] MGCV Bam error

[Ades, James]

Hi all,

In running bam(conn ~ region * timepoint + s(subjectID, bs = "re", by = timepoint) + s(subjectID, region, bs = "re", by = timepoint), data = tot.add.fil, method = "fREML", discrete=TRUE). I receive the following error.

Error in seq.default(xl[1], xl[2], length = m) :   'from' must be a finite numberIn addition: Warning message:In seq.default(xl[1], xl[2], length = m) : NAs introduced by coercion

Much thanks!

James

Googling this error will lead to warnings regarding the timepoint dummy coding (not explitly linked with multilevel modeling of mgcv). I double-checked timepoint in the dataset (using the same provided sample), and it seems that everything is defined correctly.

	[[alternative HTML version deleted]]