[P] Making KNN more efficient with pre-elimination

Hi all, I've been working on a little project which I would like some completely honest feedback on. KNN can be an incredibly powerful algorithm, but unfortunately, it can also be incredibly inefficient and can be very resource intensive to run. I've been working on a project which according to some basic testing, has shown to be on average about 20% faster than normal KNN, with little to no quality decrease. You can check out the github repo here:

https://github.com/jakeSteinburger/peKNN …

algorithm basic feedback knn machinelearning making project testing