Your favorite neighborhood machine-learning beer snob.

This program is our final project for our Artificial Intelligence class at WPI. We aimed to see if it was possible to classify beers by style (e.g. IPA, Pale Ale) based on different sets of attributes of that beer (e.g. abv, ibu, etc.). This project makes use of the scikit-learn library for our classifiers, and BreweryDB for our beer data.

This project uses Python 3.5. After cloning/forking the repo on to your machine, install requirements using pip in a new virtualenv or directly on to your machine: pip install -r requirements.txt

After the environment is setup, you must download the beer data. You will need a premium API key from BreweryDB to continue. Export your API key as BEER_API_KEY to an environment variable. Then run download_beer_data.py (note this could take a long time, there is a lot of data!): python download_beer_data.py

Now you need to transform the json data into a format usable by the classifiers. To do this, run data_transformer.py: python data_transformer.py

To run the program, simply call python main.py [trim_level], where trim_level is optional. You may enable cross-validation, change the attributes for the classifiers to learn with (trim_level), and change the amount of train/test data via variables in the main.py file.