Knowledge Maxxing← Back to topic library
ML MaxxingIntermediate75 sec
Clustering
Clustering groups examples into sets based on similarity, without any predefined labels for what each group represents.
Prerequisite: Unsupervised learning
What matters
Because clustering is unsupervised, the meaning of each discovered group has to be interpreted afterward by a person; the algorithm only finds structure, not labels.
Example
Clustering customers by purchase behavior might reveal a group of frequent small-purchase shoppers, a label a person assigns after examining the cluster.
Common misconception
Clustering does not automatically tell you what each group means or how many groups there 'should' be; both require human judgment or additional analysis.
Recall question
Why does a person need to interpret a cluster's meaning after clustering runs?
Keywords and related concepts
clusteringgroupingunsupervised learningk-means