WatchCBB

An app to provide personalized recommendations of entertaining college basketball games to watch, built from scratch over 3 weeks as my Insight Data Science project. Find it at watchcbb.me.

Contents

• The data and notebooks directories have their own READMEs with descriptions of the contents.
• scripts contains various utility scripts for scraping/cleaning data.
• watchcbb is a python module containing all of the "meat" of the app
  • watchcbb.scrape: submodule containing classes for scraping sports-reference/reddit/ESPN.
  • watchcbb.flask: submodule containing the flask web application
  • watchcbb.utils: utility functions for dealing with game data. Compile into season stats, add composite metrics, split into train/test by year, etc.
  • watchcbb.efficiency: function for computing SoS-adjusted efficiencies and other advanced stats
  • watchcbb.sql: utility class for interacting with PostgreSQL database.
  • watchcbb.reddit_utils: utility functions for managing/parsing reddit gamethread data
  • watchcbb.teams: class/functions for handling team data
• test contains all of the unit testing framework
• setup.sh: add watchcbb to your python path
• run_flask.py: launch the flask web app contained in watchcbb/flask
• ec2_setup.sh: installation commands and many helpful comments for setting up an environment on an AWS EC2 instance.