applied statistics terms

Question 1

what is clustering about?

Answer 1

finding discrete groups with small differences between group members

Question 2

is clustering classification?

Answer 2

no

Question 3

hard clustering

Answer 3

each data point is only assigned to a single cluster

Question 4

soft clustering

Answer 4

a datapoint is assigned with a certain degree of strenght over all clusters (not used in this course)

Question 5

can you add data to K-means clustering?

Answer 5

yes, by adding it to a cluster that already exists

Question 6

what method pre-specifies number of clusters?

Answer 6

K-means clustering

Question 7

what method creates a dendogram?

Answer 7

hierarchical clustering

Question 8

how does k-means clustering work?

Answer 8

you specify how many clusters you want (k) and it creaters that many random centers. it adds the closest data point to those clusters.

Question 9

what is agglomerative clustering?

Answer 9

from the bottom to the top, focusses on mergers

Question 10

what’s divisive clustering?

Answer 10

from top to bottom, focusses on splits

Question 11

how to calculate binary distance after clustering for binary data?

Answer 11

Jaccard distance, (intersection / union) or manhattan

Question 12

how to calculate binary distance after clustering for continuous data?

Answer 12

Euclidian or manhattan

Question 13

which distance techniques look at absolute distances?

Answer 13

Euclidian and Manhattan

Question 14

which distance techniques look at relative distances?

Answer 14

Jaccard and Bray Curtis

Question 15

what is linkage?

Answer 15

calculating the distance between (sub)clusters in hierarchical clustering

Question 16

single linkage

Answer 16

shortest distance, two closest points in the two clusters

Question 17

complete linkage

Answer 17

longest distance, farthest points in the two clusters

Question 18

centroid linkage

Answer 18

distance between the centroids of the two clusters

Question 19

average linkage

Answer 19

average distance of all the pairwise distances

Question 20

Wards minimun variance method linkage

Answer 20

you compute a centroid for the two clusters if they were merged (so one centroid for both clusters) and then computes the distance of the new centroid to all datapoints, and looks at the sum of squares for all compared to the new centroid

Question 21

what linkage to use when you want to do anova or regression?

Answer 21

Ward

Question 22

how does single linkage shape clusters?

Answer 22

it elongates

Question 23

how does complete linkage shape clusters?

Answer 23

they become compact

Question 24

what is the within cluster sum of squares if all data points are its own cluster?

Answer 24

0

Question 25

what is measured when all data points belong to one cluster and you measure sum of squares?

Answer 25

the variance

Question 26

what is WSS?

Answer 26

Sum of squares

Question 27

what number of clusters to choose?

Answer 27

the one were adding one doesn’t improve the total sum of squares much (the elbow)

Question 28

how to test the quality of a clustering?

Answer 28

silhouette score

Question 29

when do you have a better silhoutte score?

Answer 29

when the within cluster distance is smaller than the between cluster distance

Question 30

what is better, higher or lower silhouette score?

Answer 30

higher

Question 31

what is the difference between Manhattan and Jaccard distance?

Answer 31

Manhattan looks at absolute distance, Jaccard at relative

Question 32

what is a tanglegram?

Answer 32

a way to compare two dendograms

Question 33

what does the clustering coefficient say?

Answer 33

the degree of clustering, how related points are