NBA games : data collection & model creation

This project is currently on stand by.

This repo is centered about NBA games. The main goal here is to create a model that will predict a winner for a NBA game.

There are differents steps for this project :

1. Collect data
2. Format data
3. Create model
4. Automatize data collection
5. Predict on next games (real)

At this moment I'm at [step 3].

1. Collect data

You can find how I collect the data on the scripts/ folder.

I would like to thanks nba stats website which allows all NBA data freely open to everyone and with a great api endpoint.

• get_games.py : collect all games from yesterday to 2003
• get_teams.py : collect all teams from NBA
• get_players.py : collect all players from NBA based on games dataset [WORK IN PROGRESS]
• get_game_stats.py : collect games details based on games dataset

Also this is the script that will get all new games (but you need old datasets available on Kaggle here : dataset link) and don't forget to put it in data folder and to indicate it into the script :

• get_new_games.py

2. Format data

You can find the script that format data into scripts foled :

• format_games_for_model.py : format dataset for the model

3. Create model

[WORK IN PROGRESS]

4. Automatize data collection

5. Predict on next games (real)

Thanks for reading, Nathan