Collaborative Filtering

Collaborative filtering recommends things by looking at people with tastes like yours.

What it does

Collaborative filtering recommends things by looking at people with tastes like yours.

Neighbor: Someone with similar taste • Preference: What a user likes

How it works

• Look at what you liked before.
• Find other people who liked similar things.
• Suggest things those people liked that you have not tried yet.

Uses and applications

• Netflix can recommend shows liked by people with similar taste.
• Spotify can suggest songs that similar listeners enjoyed.
• An online shop can suggest items bought by people who bought similar products.

In short

It is like trusting a friend with similar taste.

Related methods

• KNN (Neighbors): Shares the idea of learning from the most similar examples.
• Apriori Algorithm: Finds items bought together, a different route to recommendations.
• Multi-Armed Bandit: Balances showing proven items with trying new ones.

Back to Patterns, anomalies and recommendations · All topics

The interactive demonstration and the Python walkthrough for this topic need JavaScript.