yf-ivanguo / OctaPicks

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Create W/L Features (Women)

yf-ivanguo opened this issue · comments

Feature:
(Women) {Fighter A, Fighter B} x {wins, losses, draws} x {KO/TKO, submissions, decision, overall} x {Strawweight, Flyweight, Bantamweight, Overall} x {year, all-time}

Should end up being 72 extra features.

Create a new empty CSV file and create these new features by transforming ufc_men_stats_by_fight.csv
This will be a time-consuming operation - make sure to utilize vectorization to speed up the testing process.

Vectorization:
https://pythonspeed.com/articles/pandas-vectorization/

Pandas Apply (For use when vectorization is not possible):
https://pandas.pydata.org/docs/reference/api/pandas.DataFrame.apply.html

Example (__create_home_adv_feats() uses apply, which is slower than __create_surface_env_feats() that uses vectorization):
https://github.com/ivan-guo/TennisModel/blob/main/processing.py