>>185443, >>185444, >>185445, >>185446, >>185447, >>185448, >>185449, >>185450, >>185451, >>185452, >>185453, >>185454, >>185455, >>185456
DataRepublican (small r) @DataRepublican - The short version is that it sorted on similarity and went through a recursive cache tree that exploited CPU locality. (Harder than it sounds, given categorical features.) Also did some tricky C++ to skip the step which converted from disk altogether.
Conversation
EngrStudent @EngrStudent
I love this.
I love classification and regression tree ensembles, and have for over a decade.
I would be delighted to look at your content and just understand what you did.
I have my own work on distilling random forests into single cart models, but I haven’t shared it. I really don’t know anybody who could understand it or value it.
https://x.com/DataRepublican/status/2063652983733473282

DataRepublican (small r) @DataRepublican - Challenge accepted, researching your chamber now
Conversation
James Surowiecki @JamesSurowiecki
Why do you bother writing stupid shit like this? It just makes you look like a fool to everyone outside your echo chamber.
https://x.com/DataRepublican/status/2063650418962604504

DataRepublican (small r) @DataRepublican - Probably by tonight or tomorrow night