Clustering and Association
Unsupervised learning splits in two as well. One of them puts similar things into groups. The other notices which things keep turning up together.
1Clustering
Give a model a pile of pictures of birds and animals with no class labels at all, and it still comes back with two groups — all the animals in one, all the birds in the other — formed purely on the similarity of their characteristics.
Switch between Clustering and Association, then look up the branch: neither of them is given labels.
Full lineage: AI Models → Learning Based → Machine Learning → Unsupervised → Clustering
Clustering finds the similarities itself and invents the groups. It is given no classes at all.
That is why classification is supervised and clustering is unsupervised.
The only features the model knew were tempo and intensity. The clusters of “likes” and “dislikes” were never given to it; it grouped them itself. OTT platforms recommend this way.
2Association
Look at what customers actually bought. Customer A bought bread and butter. Customer B bought bread and butter. Now customer X buys bread — what will they most probably buy?
Butter. Not because anyone wrote that rule down, but because the purchase pattern of the other customers says so. A meaningful association like that is exactly what a shop uses to recommend items.
Two ways to look at a shopping list
Association looks at the items: bread and butter keep leaving the shop in the same bag, so they belong together.
3All four sub-categories
This is the table worth learning by heart. Two supervised, two unsupervised.
| Model | Under | It gives you |
|---|---|---|
| Classification | Supervised | A category from a fixed set — discrete |
| Regression | Supervised | A value from a range — continuous |
| Clustering | Unsupervised | Groups it worked out for itself |
| Association | Unsupervised | A rule: whoever takes this usually takes that too |
4From the board papers
(A) Classification model (B) Regression model (C) Association model (D) Clustering modelCBSE board exam 2026
▶Show the answer
(C) Association model. The sentence in the question is itself an association rule — two items that keep appearing in the same basket. Clustering would have grouped the customers, not linked the products.
(a) A marketing company segments its customer base into distinct groups based on purchasing behaviour.
(b) A grocery store identifies which products are commonly bought together.Class X handbook, Reflection Time
▶Show the answer
(a) Clustering. Groups of similar customers, with no predefined segments.
(b) Association. Relationships between items in the same transaction.
▶Show the answer
Clustering is an unsupervised model that finds similarities between objects, places alike objects in the same cluster and separates them from objects in other clusters.
Classification uses predefined classes into which objects are assigned, and so needs labelled data — which makes it supervised. Clustering is given no classes and works them out from the data itself.
5Check yourself
Clustering is ______ learning, and its goal is to ______.
Clustering algorithms are commonly used in unsupervised learning for:
Which model finds interesting relationships between variables in a database?