A full stack application based on python and mongodb to scrape game data from https://stats.nba.com/ and build a continuous predictor.

First, clone repo if you haven't already and navigate into project directory e.g. Documents/nba_score_analyzer:

$ cd Documents/nba_score_analyzer

Initialize and activate a virtualenv:

$ virtualenv --no-site-packages env
$ source env/bin/activate

Install requirements:

pip install -r requirements.txt

Either run script with one argument to scrape a specific date:

python3 nba_scraper.py 02/27/2020

Or run script with two arguments to scrape games that fall in a specific interval:

python3 nba_scraper.py 02/12/2020 02/27/2020

The scraper will then access the endpoint and scrape the response with the formatting pictured below and store the json in as a document in the MongoDB database.

Response Header	Response Body

As of now everything concerning the prediction is located in the notebook called predictor.ipynb. The goal is to predict the winner of a future game given the features derived from past games. Therefore, I have fitted the following classifier:

• XGBClassifier
• SVC
• Logistic Regression

• For play by play

• General static scraper

• Logistic Regression Classifier

• Paper on prediction of game outcome

• Player stats from inteval

• Continual Learning