Clustering: Things you should know

I know many of you know the definition of clustering but few of you also might don’t know. So basically, clustering is a kind of grouping when we don’t have any idea how to make the groups and on what basis? In data science term it is unsupervised learning.

First, you have to understand the difference between types of clustering and types of clustering algorithms. Many of you consider the types of clustering algorithms i.e. K-means, Hierarchical as types of clustering but it’s not. The following are the types of clustering:

Yes, you saw right these are the main types of clustering. Now we will look for each type.

1.    Partitional: Partitional clustering is nothing but making subsets of data that will not overlap. In simple words making a cluster of each data object. So in partitional clustering, the number of clusters is always the number of observations.

2.    Hierarchical: If we make sub-clusters of the cluster then it would be called hierarchical clustering. It is the set of nested clusters that are organized as a tree.

3.    Exclusive: When variables make clusters of itself then that clusters are called exclusive clusters. e.g. if the dataset contains 20 variables then it will form 20 clusters itself.

4.    Overlapping: If the objects of the dataset find in more than one cluster then that clusters are considered as overlapping clusters. Overlapping clusters are also called as Non-Exclusive clusters.

5.    Fuzzy: Sometimes objects of the dataset belongs to every cluster, that time those clusters are considered as fuzzy clusters.

6.    Complete: When every object of the dataset is added in clusters then those clusters are called complete clusters.

7.    Partial: When clusters fail to add every object of the dataset. Then these clusters considered partial clusters.