A personal project that encompasses NBA data scraping, data cleaning, data transformation, and a neural network model.
Disclaimer: Data is graciously and thankfully scraped from Basketball Reference. They're the best!
The Scraping
Jupyter notebook contains all the necessary code for scraping data from Basketball Reference.
The NBASalaries
Jupyter notebook contains everything else - cleaning, transformation, and the model.
The model has the following features:
- FG: Field goals
- FGA: Field goals attempted
- FT: Free throws
- FTA: Free throws attempted
- PTS: Points
- AST: Assists
- TOV: Turnovers
- 2P: 2-pointers
- 2PA: 2-pointers attempted
- GS: Games started
- MP: Minutes played
Something that stood out to me at first, but made more sense when I thought about it was the low(-er) correlation between salary and 3-pointers. When you think about it, players like Jokic + Gobert + Embiid hardly take that many 3-pointers, but are among the highest paid NBA players.
This is the correlation matrix (2021-22 is the salary variable):