Stephan Zednik
github: https://github.com/zednis/itws6960-2017-final-project
Open Food Facts is a free, open, collaborative database of food products from around the world, with ingredients, allergens, nutrition facts and all the tidbits of information we can find on product labels.
Individual contents of the database are available under the Database Contents License.
Available Formats: CSV/TSV and SQLite.
Data access link: https://www.kaggle.com/openfoodfacts/world-food-facts
Dataset description: https://world.openfoodfacts.org/data/data-fields.txt
Over 20k recipes listed by recipe rating, basic nutritional information, assigned category, ingredients, and cooking instructions lifted from http://www.epicurious.com/recipes-menus.
The kaggle dataset page does not appear to specify a license for the dataset publication.
Available Formats: JSON (full recipe data)
data access link: https://www.kaggle.com/hugodarwood/epirecipes
The application uses postgresql and mongodb for data persistence.
Instructions for installing postgresql are available at https://www.postgresql.org/docs/9.2/static/tutorial-install.html.
Instructions for install mongodb are available at https://docs.mongodb.com/manual/installation/.
The scripts/setup.sh script, and the data-loading scripts it calls, expect there to be local instances of postgresql and mongodb.
To change postgresql connection info modify the DATABASE object in scripts/food_fact_models.py.
To change mogodb connection info modify values from the run function in scripts/load-recipes.py.
- python3
- psycogp2
- sqlalchemy
- pymongo
- jupyter
- pandas
- matplotlib
If you have python3 installed you can install the other python dependencies by running
python3 -m pip install -r requirements.txt
from both the scripts/ and app/ directories.
The scripts/load-open-food-facts.py and scripts/load-recipes.py scripts will load the data into a locally-running postgresql and mongodb database, respectively.
These scripts are called by scripts/setup.sh and need not be run individually.
The application is a jupyter python3 notebook. This notebook loads libraries that provide fluent APIs for querying the product and recipes datasets and includes a locally-defined class DistributedQuery which implements a fluent API that allows for federated queries against both databases.
jupyter should be installed by running
cd app
python3 -m pip install -r requirements.txt
or manually npm with command:
python3 -m pip jupyter
Running jupyter is as easy as typing
jupyter notebook
from the app directory.
More information on installing and running jupyter notebook can be found at http://jupyter.org/install.html.
After starting jupyter go to URL printed to the commandline (e.g. localhost:8888) and open the app.ipynb notebook.
Instructions and examples for using the product and recipe APIs for querying data are available in the notebook.
More information about running jupyter cat be found at https://jupyter.readthedocs.io/en/latest/running.html#running.