Using Random Forest Classification to see if NBA Positions in the years 2016-2020 could be determined solely from box score statistics. We find that after tuning max_depth to allow for more complicated trees, we can achieve an accuracy of 65%, with Point Guards and Centers being the most easily distinguishable.
Data sourced from Basketball-reference.com