You are on page 1of 15

Introduction to Data Anlytics

(ITEC 3040)

Clustering
(Modified from Aijun Ans slides)

Outline

1
Density-based Clustering

Density-based Clustering

2
DBSCAN: Basic Concepts

DBSCAN: Basic Concepts

3
DBSCAN: Core, Border and Noise Points

DBSCAN: Input Parameters

4
DBSCAN: Basic Concepts

DBSCAN: Basic Concepts

10

5
DBSCAN: Cluster

11

DBSCAN: The Algorithm

12

6
Another Version of DBSCAN
(for simplicity, not efficiency)

13

DBSCAN: Core, Border and Noise


Points

14

7
When DBSCAN Works Well

15

When DBSCAN Does NOT Work


Well

16

8
DBSCAN: Sensitive to Parameters

17

Comparing DBSCAN and K-means

18

9
Comparing DBSCAN and K-means (Contd)

19

DBSCAN: Determining Eps and


MinPts

20

10
Outline

21

Cluster Validity

22

11
Different Aspects of Cluster
Validation

23

Clusters found in Random Data

24

12
Measures of Cluster Validity

25

HA Index: an External Index

26

13
SSE: an Internal Index

27

DB: an Internal Index

28

14
Final Comment on Cluster Validity

29

Summary

30

15

You might also like