Try the Kaggle competition

[vingkan]

1. Join the Kaggle competition at this link.
2. Fork Vinesh's competition demo notebook.
3. Run through the notebook. It (1) helps load the data, (2) provides some working recommenders, and (3) saves your predictions for the Kaggle leaderboard.
4. Experiment with different recommender systems. The module redcarpet provides implementations of collaborative, content-based, and hybrid recommenders. Try different parameters, similarity functions, and hybridizations. Refer to the docstrings for each method by calling help(method_name).
5. Choose your best model and use the last section of the notebook to save the predictions. This will make predictions for the test data (for the public leaderboard) and the hold out set (for the private leaderboard). You do not have access to the hidden entries of the hold out set.
6. Click the "commit" button to run your notebook from top to bottom to reproduce your work. If there are errors in any cell, you will have to fix them and try to commit again.
7. After the commit finishes, click the "open version" button to view your kernel. Click on the output tab and download the .csv file of your predictions.
8. Go to the Kaggle leaderboard and submit your .csv of predictions.
9. Now you should be on the leaderboard!