Question

1. You are given the follow information: You need to apply k-means clustering,Your dataset has 1,000 observations, Your dataset has 57 features, K=2
2. Answer the following questions
   1. How are the initial centroids selected?
   2. How many clusters will be produced?
   3. What measure is used to evaluate the quality of the clusters?
   4. For the evaluation measure, do higher or lower values indicate better clusters? Why?

Answer 1

In the question following information is given...

N = 1000 observations

d = 57 features

K = 2 (number of clusters)

Solution :-

i. Initial centroids are selected at random.

ii. 2 number of clusters will be produced. (Because K = 2).

iii. Quality of the clusters can be measures by distance measure (Euclidean distance) known as L2 distance.

iv. For the evaluation measure, lower values indicate better clusters. The distance can be defined as follows :

Our objective is to chose cluster center in such a way that within cluster distance is minimized from the centroid of the cluster.