How Does StatQuest: Hierarchical Clustering Work?

StatQuest shows how hierarchical clustering takes a bunch of things and groups them into bigger and bigger families, like when you sort your toys by color or size.

Imagine you have a pile of different fruits: apples, bananas, oranges, grapes, and kiwis. You want to group them based on how similar they are, maybe by size or taste. Hierarchical clustering is like slowly putting together a family tree for these fruits.

It begins with each fruit being its own little group. Then it finds the two fruits that are most alike, say, two grapes, and puts them in a pair. That’s like when you pair up for a game at recess.

Next, it looks at all the pairs and sees which ones are most similar. Maybe the two grapes are close to another grape, so now they form a small group of three grapes. This keeps happening: the closest groups get joined together, like adding more friends to your team.

Eventually, all the fruits become one big family tree, from smallest pairings up to everything being grouped together. That’s how hierarchical clustering works, it builds bigger and bigger families by always joining the most similar ones first!