I really enjoy watching project walkthrough videos posted by Dataquest on its YouTube channel. The one about building book recommendation systems using collaborative filtering was especially interesting to me.
I decided to implement my own project along similar lines, but added some tweaks that I hope made the workflow more robust and the resulting recommendations more relevant. I used the same UCSD Book Graph datasets to build a simple recommendation system that produced several dozen recommended books for me.
The link to the project on Github is here. I plan to continue working on this project by adding machine learning algorithms and using validation to evaluate the errors of predictions.
The Jupyter notebook can be downloaded here (126.1 KB).
I will be grateful for your feedback.
Click here to view the jupyter notebook file in a new tab