The link to the gapanalysis script is to Bitbucket, and apparently the authoritative location is now on GitHub: https://github.com/haukex/htools/blob/master/gapanalysis.

For obvious reasons, I'd really like to see this capability enter the PDL-verse. An obvious location would be PDL::Stats::TS. If you agree, want to create an issue on the repo for that, or even better, a PR?

Even without the above, the test data here could be generated more easily using PDL::GSL::RNG's ran_shuffle_vec and PDL's native random stuff.