| Detailed description |
I'm actually using TraMineR 2.0-5, but this feature has been around since I started using TraMineR in 2015.
I believe (I'm not a programmer) that seqdist stores distance data in a 2-dimensional matrix, where the index is 1-dimensional and 32bit. It's a feature in R, which I think could be circumvented.
library(TraMineR)
seqdata <- dplyr::tibble(t1 = sample(1:3, 46342, replace = TRUE)) # sample data stored in a tibble
seq <- seqdef(seqdata)
dist <- seqdist(seq, method = "HAM", full.matrix = FALSE)
...returns
Error in seqdist(seq, method = "HAM", full.matrix = FALSE) :
negative length vectors are not allowed
If you repeat the code with one less case/row, it will run.
One option might be to look into long vectors https://stat.ethz.ch/R-manual/R-devel/library/base/html/LongVectors.html
|
|
