Clustering

Clustering is an unsupervised learning technique. It is used for grouping data points, or objects that are somehow similar. Clustering means finding clusters in a dataset, unsupervised. Clustering is one of the most used applications of unsupervised learning.

Motivation

Generally, clustering can be used for one of the following purposes:

• Exploratory data analysis
• Summary generation
• Outlier detection[1]
• Finding duplicates
• Pre-processing step

Types pof clustering

Some divide clustering into two subgroups:

• Hard clustering: Each data point either belongs to a cluster completely or not. Clusters do not overlap.
• Soft clustering: A probability or likelihood is assigned for putting data points into separate clusters. Clusters may overlap.

Algorithms

Some of the commonly used clustering algorithms are:

• Partitioned-baseed clustering:
  • K-means:
  • K-median:
  • Fuzzy c-means:
• Hierarchical clustering:
  • Agglomerative clustering:
  • Divisive clustering:
• Density-based clustering:
  • DBSCAN:

others

• Expectation maximization:
• Hierarchical cluster analysis (HCA):

External links

• From Hard to Soft Clustering Pavel A. Pevzner

References

• [2]
• [3]
• [4]