Scribd
Upload
18 views

Hierarchical Clustering.pptx

The document discusses hierarchical clustering, detailing two main types: divisive and agglomerative clustering. Divisive clustering follows a top-down approach, starting with one large cluster and splitting it into smaller ones, while agglomerative clustering begins with individual observations and merges them into larger clusters. It also explains how to calculate similarity between clusters using methods like single linkage, complete linkage, and average linkage.

Uploaded by

Asma Ayub
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
18 views

Hierarchical Clustering.pptx

The document discusses hierarchical clustering, detailing two main types: divisive and agglomerative clustering. Divisive clustering follows a top-down approach, starting with one large cluster and splitting it into smaller ones, while agglomerative clustering begins with individual observations and merges them into larger clusters. It also explains how to calculate similarity between clusters using methods like single linkage, complete linkage, and average linkage.

Uploaded by

Asma Ayub
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 96

Hierarchical Clustering

TYPES OF CLUSTERING
TYPES OF HIERARCHICAL CLUSTERING
Divisive Clustering
• The approach is Top – Down.
• In divisive clustering, we start from a situation where all observations
are in a same cluster e.g. Dinosaurs.
• Then we split the big cluster into smaller ones.
• We continue this procedure until each observation is in a separate
cluster.
• However, to find the best split, we must explore all possibilities at
each step.
Agglomerative Clustering
• The approach is bottom-up.
• We start from different dog and cat breeds, cluster them into cats and
dogs respectively.
• Then we continue pairing up species, until we reach the main animal
cluster.
• Agglomerative and divisive clustering should end up in similar results.
Agglomerative and divisive clustering
Agglomerative and divisive clustering
Agglomerative Hierarchical Clustering
• We start with each case being its own cluster.

• Using some similarity measure like Euclidean Distance, we group the


two most similar clusters together, reaching N-1 cluster solution.
Agglomerative Hierarchical Clustering
• Then we repeat this procedure until all observations are in a single
cluster.
• This is called a Dendrogram.
Example of a Dendrogram
Dendrogram
• At the start, each country is a separate cluster.
• The two countries that are merged together are Germany and France
based on the features considered.
• Going up, Germany and France will be considered as one cluster.
• Going up, all the countries end up in being one big cluster.
• The bigger the distance between two links, the bigger the diiference
in terms of the features.
Number of Clusters
• Two clusters in this case

• Three clusters in this case


Number of Clusters
• How to decide, where to draw the line.
• There is no rule of thumb
• After solving many problems, an intuition can be developed.
• When the distance between the two stages is too big, it’s a good idea
to stop there.
Calculating similarity between two clusters
• Single Linkage
• Complete Linkage
• Centroid Linkage
• Average Linkage

• Depending upon the dataset, method is selected.


Calculating similarity between two clusters
Single Linkage
• The distance between two most similar parts of the cluster is taken.
Single Linkage
Single Link
Single Link
Single Link
Single Link
Single Link
• Based on the minimum distance
Single Link
Single Link
Single Link
Single Link
Single Link
Single Link
Single Link
Single Link
Single Link
Single Link
Single Link
Single Link
Single Link
Single Link
Single Link
Complete Linkage

• The distance between two least similar bits of cluster is taken. i.e., the
farthest point in the cluster.
Complete Linkage
Complete Linkage
Complete Linkage
Complete Linkage
Complete Linkage
Average Linkage
Pros and Cons

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy