Item-based or user-based

Building Recommendation Engines in Python

Rob O'Callaghan

Director of Data

Item-based filtering

Building Recommendation Engines in Python

User-based filtering

Building Recommendation Engines in Python

Why use item-based filtering?

Pros:

  • Item-based recommendations are more consistent over time

  • Item-based recommendations can be easier to explain

  • Item-based recommendations can be pre-calculated

Cons:

  • Item-based recommendations result in very obvious suggestions
Building Recommendation Engines in Python

Why use user-based filtering?

Pros:

  • User-based recommendations can create a lot more interesting suggestions

Cons:

  • Generally beaten by item-based recommendations using standard metrics
Building Recommendation Engines in Python

Let's practice!

Building Recommendation Engines in Python

Preparing Video For Download...