Python machine learning model that receives a set of scores (usually top 100) from some Osu! map (Osu! is a rhythm game) and calculates each score's uniqueness compared to its peers.
The calculation is based on the mods used, max combo, pp earned, and accuracy.
This is by no means an attempt to define good scores, and "uniqueness" is hard to define, as what is unique may differ from each person, but I thought this would be a good start and good opportunity for finding unique scores.