NBA Deep Learning

How to get the data

Running python preprocess.py should create all the data required to run the model. It will take a few minutes to run. It takes all the boxscores in games2017 and games2018 and synthesizes it into our data format.

Note: You may need to make a folder called "final_data". While testing, we have found that on some systems, python can create the directory, while on others, it will throw an error if the directory does not exist.

How to run the model

Run python model_framework.py --[model option] to run the model.

• The flag --dense uses the dense model.
• The flag --team_conv uses the team level convolution model
• the flag --player_conv uses the player level convolution model