Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Machine Learning the 2015-16 NBA All Stars (The Story of a Rookie Dev)
5 points by fatalfame on Feb 11, 2016 | hide | past | favorite | 2 comments
http://github.com/fatalfame/nba_machine_learning

I wrote a python script to machine learn data from historical nba stats and attempt to predict this year's all star roster. Links to my results can be found in the readme file on git. I welcome any feedback or contributions. Or if you just want to take a look, enjoy!



That's neat! You should add team winning percentage as another stat to see if it improves your results. It might reduce the occurrence of false positives for players like Lillard and Rondo.


Just need to make sure it's not weighted enough to remove the All Stars coming from worse (<= .500) teams (Carmelo, Drummond, Wall, Kobe, Harden, Davis, Cousins).

On a side note, I'd say the reason Lillard (I can't speak for Rondo) wasn't selected is more related to the over-abundance of top tier guards in the West this year, rather than the Blazers' win percentage. He's the only player in the top 10 PPG that isn't an All Star, and he's been leading Portland to a much better year than pretty much anyone predicted.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: