Barry G Becker

I will look into 6 some more. I think that is the one I have seen fail intermittently as well. Sometimes I see it give the same result many times...

I just ran the tests 7 times in a row. They all passed except for the second time when I got [info] - Run MDLPD on all columns in titanic2...

I just ran the tests 10 times in a row after pulling and building in my Ubuntu linux VM and there were no failures at all. I have also had...

I think this issue still persists - though it is hard to reproduce. I don't know where the source of the non-determinism is, but I think it is premature to...

I really think that this issue should be reopened. Sometimes inserting print statements in the code can cause it to run at different rates and give different splits.

I will need to take a closer look at this soon. I just got a new machine and tried running the MDLP unit tests and found that one of them...

I haven't solved this yet, but I think I have narrowed it down. I made some local changes so that findSmallThresholds was called for all features. I noticed that I...

I added some unit tests for the Few and ManyValuesThresholdFinders in my fork. Though I am seeing intermittent failures in the other tests, the tests I added for the Threshold...

I was reading about an [interesting case in scala where parallel collections can lead to non-deterministic results](http://docs.scala-lang.org/overviews/parallel-collections/overview.html) if a variable declared outside a closure is accumulated in the closure. ```...

Another possibility might be in the initialThresholds method, but currently don't have any theories as to the source. I see the intermittent failures fairly often, but it is not reliably...