Unsupervised learning

Question 1

What is unsupervised learning

Answer 1

collects similar data into clusters (they get names cluster A etc but this has no semantic meaning)

Question 2

What is the process of clustering

Answer 2

grouping similar objects into groups

Question 3

What are example applications of clustering

Answer 3

social network analysis or marketing
Image segmentation

Question 4

What are some unsupervised learning algorithms

Answer 4

clustering concepts
Partition-based clustering algorithms (k-means)
Hierarchical clustering (agglomerative clustering)

Question 5

What is the aim of clustering algorithms

Answer 5

to see whether the data fall into distinct groups, with members within each
group being similar to other members in that group but different from members of
other groups

Question 6

What are the steps of the k-means algorithm

Answer 6

1. 2. define the number of clusters (k)
      choose k data objects randomly to serve as the initial centroids for
      the k clusters
2. assign each data object to the cluster represented by its nearest
   centroid
3. find a new centroid for each cluster by calculating the mean vector
   of its members
4. undo the memberships of all data objects; go back to step 3 and
   repeat the process until cluster membership no longer changes or a
   maximum number of iterations is reached

Question 7

What are k-means variations

Answer 7

• selection of the initial k means
• dissimilarity calculations
• strategies to calculate cluster means
• using different distance measures

Question 8

What are the k-means stengths

Answer 8

simple and easy to implement
quite efficient

Question 9

What are the weaknesses with kmeans

Answer 9

• Need to specify the value of k, but we may not know what the value should be beforehand
• you may want to experiment with k value (i.e. elbow)
• Sensitive to the initialisation
• sensitive to noise
• clustering of different sizes

Question 10

Why is k-means sensitive to the noise

Answer 10

since we are using the mean

Question 11

Explain hte agglomerative clustering algorithm

Answer 11

1. take all n data objects as individual clusters and build a n x n
   dissimilarity matrix storing distances between any pair of data objects
2. while the number of clusters > 1 do:
3. 2. 3. find a pair of data objects/clusters with the minimum distance
         merge the two data objects/clusters into a bigger cluster
         replace the entries in the matrix for the original clusters or
         objects by the cluster tag of the newly formed cluster
4. re-calculate relevant distances and update the matrix
   * whole process produces a dendrogram
   * relies on the definition of a distance metric between clusters

Question 12

WHat are the strengths of agglomerative clustering

Answer 12

• deterministic results
• multiple possible versions of clustering
• no need to specify the value of a k before hand
• can create clusters of arbrary shales

Question 13

WHat are the weakensses of agglomerative clustering

Answer 13

does not scale up for large data sets