A Survey of Parallel Clustering Algorithms Based on Spark

Table 1

Comparison of key features of various algorithms.

	Clusters shape	Handling outliers	Input parameters	Data points’ distribution
K-means	Spherality	No	m	Random
			n
Hierarchy	Spherality	No	m	Random or by data space
Density	Arbitrary	Yes	Eps	Data space
			MinPts
Model	Mathematical model	Yes	m	Random

m: number of clusters; n: maximum number of iterations; Eps: radius of data point neighborhood; MinPts: minimum density of core data point; : parameters required for a specific model.