Unsupervised Learning

Question 1

Learning method where TARGET LABELS are unknown

Tries to find the innate structure of data
Eg clustering

Answer 1

Unsupervised learning

Question 2

Groups the data into K CLUSTERS based on data’s DISTANCE FROM EACH CLUSTER’S CENTER

Answer 2

K means clustering

Question 3

Each cluster’s centroid is considered a

Answer 3

Centroid

Question 4

What is the algorithm of K Means clustering

Answer 4

> initialize k centroids RANDOMLY
for each data, compute distance from each centroid
get average of coordinates of data assigned to each centroid
if centroid stay the same, stop, else compute distances again

Question 5

What do you do if a centroid has no data point?

Answer 5

Restrat the algorithm and choose different centroids

Question 6

True or false: kmeans always produces the optimal solution

Answer 6

False. It can get stuck to a LOCAL optimum especially if value of k is TOO SMALL