KMeans

Question 1

Good clustering will produce clusters with:

• [High / Low] intra-class similarity.
• [High / Low] inter-class similarity.

Answer 1

High

Low

Question 2

Other distance measures used in Clustering include:

Answer 2

Minkowski distance
Pearson correlation distance
Spearman correlation distance
Kendall correlation distance.

Question 3

Challenges with k-Means Clustering

k-Means is very sensitive to the initial randomly chosen cluster centers (this is known as the ________)

Answer 3

random initialization trap

Question 4

The _______ initialization approach mitigates the effects of the random initialization trap.

Answer 4

K-means++

Question 5

Methods for choosing the right K include:

Answer 5

Elbow Method
Information Criterion Approach
Silhouette method
Jump method
Gap statistic

Question 6

WCSS stands for ________ and is associated with the _____ method for choosing K

Answer 6

Within Cluster Sum of Squares

Elbow Method

Question 7

Strengths of K-Means Clustering?

Answer 7

- Uses simple non-statistical
principles.
- Very flexible and malleable
algorithm.
- Wide set of real-world
applications.

Question 8

Weaknesses of K-Means Clustering?

Answer 8

• Simplistic algorithm.
• Relies on chance (initial k centroids)
• Sometimes requires some domain knowledge in
  determining the ideal number of clusters.
• Not ideal for non-spherical clusters.
• Works with numeric data only.