Hierarchical Clustering
1. Introduction
계측적 군집화는 비 지도 학습 알고리즘 이다.
클러스터의 수를 사전에 정의할 필요 없다.
결과물로 dendrogram라는 트리 다이어그램을 생성한다.
2. Key Terms
cluster : 유사도기반으로 모여진 점들
Cluster Centroid : 클러스터의 평균(mean)
Distance measure : 거리 측정 척도 (Manhattan or L1, Euclidian or L2 distances, etc.)
linkage criteria : 거리계산에 사용된 위치
- Single Linkage - 가장 가까운 것에서 계산
distance is computed between the two MOST similar parts of clusters (two closest points).- Single linkage suffers from chaining meaning that clusters can be too spread out, and not compact enough.
- Complete Linkage - 가장 먼곳에서 계산
distance is computed between the two LEAST similar parts of clusters (two most distant points).- Complete linkage avoids chaining but suffers from crowding meaning that Clusters are compact, but not far enough apart.
- Average Linkage - 중앙에서 부터 계산
distance is computed between clusters’ centroids.- This is a balanced approach: clusters tend to be relatively compact and relatively far apart.
