Clustering Algorithm

Clustering:   Clustering is an unsupervised learning algorithm. It group data on similarity to each other. It finds simpler data sets. It provides categorical class labels. It is used in retail marketing, banking, insurance, publication, medicine, biology and many other fields.

We need clustering because of the following reasons:

(i) Exploratory data analysis

(ii) Summary generation

(iii) Outlier detection

(iv) Finding duplicates

(v) Pre-processing step

The algorithms which are generally used in clustering are Partition based, Hierarchical and density-based clustering.

1. Partition based clustering: It is a relatively efficient algorithm. Some partition-based clustering algorithms are k-mean, k-median, and fuzzy c-mean.

2. Hierarchical clustering:  It produces a tree of clustering. Agglomerative and Divisive are some popular hierarchical algorithms.

3. Density-based clustering: It produces arbitrary shaped clusters. DBSCAN is a popular density-based clustering algorithm.