A collaborative-based filtering recommendation system.
Tom Lafferty: https://github.com/TomLafferty
Efe Cantekin: https://github.com/Efe-C-3
Christine Roethel: https://github.com/croethel
Book recommendation system that is able to take 6 different user inputs based on user ratings of book categories and returns a machine learning based recommendation of 20 books identified by similiar existing users behaviors.
Additionally, the book recommendation system can currently search pre-existing users (based on User ID) and render 10 book recommendations based on what similiar users have also reviewed.
Currently deployed to Heroku, recommendations are currently not functional due to Heroku memory limitations.
Existing code can be successfully ran locally (note: data set is currently connected to S3 bucket, would need to configure dotenv library and .env directory locally)
Maintanance version deployed to:
https://bookrecommender22.herokuapp.com/
Python
Django
Postgresql (when deployed)
SQLite3 (when ran local)
AWS S3
TensorFlow
Scikit-learn
Heroku
Pandas
Book recommender can be installed via Pip. To start, clone the repo to your local computer and change into the proper directory.
$ python3 -m venv bookrec
$ source bookrec/bin/activate
(bookrec) $ pip install -r requirements.txt
(bookrec) $ python manage.py makemigrations
(bookrec) $ python manage.py migrate
(bookrec) $ python manage.py createsuperuser
(bookrec) $ python manage.py runserver
# Load the site at http://127.0.0.1:8000
https://www.kaggle.com/jirakst/bookcrossing
(Adjusted column names 'User-ID', 'Book-Rating' and 'Book-Title' to 'UserID', 'BookRating' and 'BookTitle' in all .csv files)
Initial:
Current:
