Confusion about clustering

What is the difference between clustering and using supervised learning to recommend something? I got confused because in supervised Andrew gave the example of recommending similar movies and in unsupervised clustering, the example of Google News matching words to make the relevant articles appear

The basic difference is that in supervised learning you have examples that are labelled. However in clustering, you are actually looking for patterns from your data.
The use-case defines the algorithm you use.

1 Like

but i think that the 2 examples are similar case use

If you look at the examples closely, the two are a bit different. Matching words to make relevant articles appear is more of clustering as it requires looking for patterns in your data. The data is not labelled for you. However, in recommending similar movies, especially if you use content-based filtering, the data is labelled with, for instance, whether the user liked the movie or not, and then the machine learning model is trained to make the prediction.
Unsupervised clustering only comes in when you group similar things in clusters and then apply a recommender system.


Clustering just lets you group the examples by similarity.
You aren’t creating a model, you’re just looking to see what the characteristics of the data are.

Supervised learning creates a model that lets you make predictions on new data.


In the example of supervised learning, the prerequisite for making a recommendation is acquiring knowledge of the user’s past behaviors, such as the videos they’ve interacted with. This implies that we have access to both the inputs (the user’s actions) and the outputs (the system’s responses), which the supervised learning algorithms utilize for making future recommendations.

On the other hand, clustering is fundamentally about identifying patterns. A prime example of this is how Google News aggregates articles based on specific keywords or topics. This method doesn’t necessarily require previous user interaction data but rather groups similar items based on inherent characteristics.

Therefore, while both supervised learning and clustering are methods used in machine learning, their underlying approaches and applications are distinct when examined in detail.

1 Like

I got u, but I think the results got from supervised and unsupervised are called models

Got it, thanks

Got it, thank u