02·Hierarchical clustering

Introduction

Hierarchical clustering builds a similarity tree (dendrogram) instead of a single flat partition. Unlike k-means it does not require the number of clusters upfront — you pick it after inspecting the tree by cutting it at a chosen height. In this lesson you will separate the agglomerative (bottom-up) approach from the divisive (top-down) one, compare the four linkage criteria (single, complete, average, Ward), understand why their cost is O(n² log n) or O(n³), and when they still beat k-means — chiefly when the data structure is naturally nested.