Chapter 9: Cluster Analysis

Question 1

what is cluster analysis

Answer 1

unsupervised

task of grouping a set of objects so objects in the same group are similar

Question 2

give the 3 similarity measures

Answer 2

Euclidean distance

cosine similarity

minkowski distance

Question 3

what are the key tasks of cluster analysis

Answer 3

define distance measure

identify cluster number

perform grouping

evaluate

Question 4

what are the hyperparameters in cluster analysis

Answer 4

distance measure

cluster number

Question 5

what is k means clustering

Answer 5

a clustering algorithm that calculates the distances to centre points

assigns to nearest

updates centre using average of cluster points

Question 6

within each cluster, what is minimised

Answer 6

sum of squares

Question 7

what is the running time of k means clustering

Answer 7

O( T K N)

Question 8

what are the drawbacks of k means clustering (3)

Answer 8

doesn’t cope well with noise or outliers

need to decide number of clusters

not suitable for complex patterns

Question 9

what does the distance between clusters tell us

Answer 9

the similarity between two points

Question 10

what is single link measure

Answer 10

distance between clusters = minimum distance

Question 11

what is multi link measure

Answer 11

distance between clusters = maximum distance

Question 12

what is average link measure

Answer 12

distance between clusters = average distance

Question 13

describe hierarchical clustering

Answer 13

objects grouped in a tree structure

Question 14

what is agglomerative clustering

Answer 14

start with atomic clusters and merge until you get one big cluster

Question 15

what is divisive clustering

Answer 15

start as one big cluster and separate out to atomic clusters

Question 16

what is a dendrogram

Answer 16

plots data points and shows the distance when they were clustered together

Question 17

what is the lifetime of a cluster

Answer 17

difference between when created and when merged

Question 18

how to we get k clusters from hierarchical clustering

Answer 18

cut the tree

Question 19

what is cluster validation

Answer 19

check the clusters make logical sense

Question 20

what are the two methods of cluster validation

Answer 20

internal and external criteria

Question 21

what pattern of variation do we want

Answer 21

a good cluster should have a small in cluster and large incluster variation

Question 22

how do we calculate within cluster variance

Answer 22

sum for each point in cluster: distance(point, center)

Question 23

how do we calculate between cluster variance

Answer 23

sum for each cluster: no_points_in_cluster * d^2(cluster, data centre)

Question 24

what is external validation

Answer 24

validate against ground truth labels

Question 25

how do we evaluate against ground through labels

Answer 25

rand index

Question 26

describe rand index

Answer 26

compare cluster ID to class ID

agreement / disagreement table

rand = (a + d) / (total)