Ryan Curtin

Are you planning to use these traits anywhere? It's not very useful to add them if they're not actually going to get used anywhere.

> It's harmless to merge these lines, I suppose. It might be harmless for you, but it's not harmless for the maintainers. For one, it's not correct: not every kernel...

I think that it was just an unlucky test failure. Some of the tests are probabilistic and while we do our best to keep the failure probability low enough that...

At least personally, I think it would be useful to further investigate why the trees differ even when using information gain. I wonder if the optimization is exposing some other...

If you were running on reasonable data, I strongly doubt you encountered the issue in #3644. That issue has gone undetected for years because the unspeakably vast majority of datasets...

Is it possible you are running out of RAM? I can try to reproduce locally (on Linux) if are able to share a link to the dataset.

If you're running out of RAM, you could try running with fewer cores. By default OpenMP will use one core to train one tree in the forest, but if each...

Dropbox perhaps? Google drive?

Hey @SrikanthPusarla, sorry it took me so long to get back to this. It has been a busy month... I took a look into the code and data you sent....

Hey there @SrikanthPusarla, I found that your data has a particularly strange characteristic that causes a crash in some trees. I detailed the actual issue in #3644 and that PR...