Looking for your next board game?
Want another game for your quarantine game nights?
Use my board game recommender system to find your next favorite!
How did I build this?
Using item-based collaborative filtering.
I collected a data set of ~15mil ratings from 190,000 users for 10,500 games by scraping the www.boardgamegeek.com API. I created a sparse user-game matrix based on this data.
After taking care of the bias and using an SVD to decrease the rank of the user-game matrix, I computed the distances between games.
Given a game, the system returns a 100 recommendations based on the query.
Tools:
Python, scipy.sparse and scikit-learn libraries, Flask, some CSS and jQuery.
Hire me!
Made by: Kostya Timchenko
