[tests] design better tests on heuristics of bimodal distros or cluster to target

Open gph82 opened this issue 8 years ago • 1 comments

Many tests are failing because one method is supposed to find the two centers of a sampled bimodal distribution. The criterion of what constitutes having found the center should be relaxed or a more deterministc test should be devised

Mar 09 '17 23:03 gph82

adding "cluster to target" to the list of improvable tests

Mar 14 '17 11:03 gph82