Building our first simple recommender system using popularity (Non-Personalized)
In the previous video we looked at the RecSys field from an overview perspective, in this video we start looking into implementing a basic one using popularity. This is in most cases your go to solution, so specifically we look at the MovieLens dataset: https://grouplens.org/datasets/moviel...
And we use num votes, mean rating, and damped mean for building the recommender system that is non-personalized, because it is just using summary statistics. This however, is a very strong baseline to start from.
Timestamps:
0:00 - Introduction
0:55 - MovieLens dataset
3:50 - Number of votes
6:40 - Mean rating
8:24 - Damped Mean
15:15 - Comparing results
16:50 - Cleaning up code
19:45 - Ending