This app can give you the Top 5 recommendations lists of articles on globo.com per user.
This was a kaggle competition held several years ago.
See News Portal User Interactions by Globo.com for much more detailed documentation.
It is in fact a python BayesianPersonalizedRanking model.
You need to install first some python librairies:
Flask
implicit==0.7.2
scipy
numpy
pandas
tqdm
scikit_learn
As you know now, our main objective is to predict a ratings (implicit way chosen) of globo's clicks dataset.
- clicks user-item related
- ratings
I strongly recommand you to install packages from the requirements.txt.
None