Repository files navigation

The Berkeley SUMSarizer

The SUMSarizer reduces raw SUMS data into a log of "cooking events" using machine learning. Upload your CSV files, train the algorithm by manually annotating a few data sets (with a nice UI), and the SUMSarizer will do the rest.

Guaranteed to save you a month of data wrangling!

Check out thesample development environment for getting started on SUMSARIZER development.

SUMSarizer is a Python 2.7 app.

Install its requirements:

pip install -r requirements.txt

R should also be installed with the following packages:

plyrdevtoolspsplinecaToolsglmnetdevtools::install_github('jeremyrcoyle/origami')

The current config expects a Postgres database at the following URI:postgresql://sums:sums@localhost/sums

The following environment variables should be set:

APP_SETTINGS="config.DevelopmentConfig"

python run.py

There are a few different background workers to spin up for various tasks:

python worker.py

python -m tasks.ml_worker

To initialize the DB schema run:

python manage.py db upgrade

If you change a model file inmodels.py, run:

python manage.py db migrate

to generate the migration file. Inspect the migration file inmigrations/versions/...py If it looks good, run it with:

python manage.py db upgrade