A survey of parallel clustering algorithms based on vertical scaling platforms for big data

Clustering, or cluster analysis, is an important unsupervised task in machine learning that determines how the observed data naturally clusters. Many efficient traditional clustering methods based on different behaviours, such as partitioning’ hierarchical, grid, model, and density based, have been proposed in recent decades. However, clustering itself is considered an NP-hard problem, and it becomes more challenging when the clustered data is large. The classical clustering techniques cannot handle big data problems due to their large volume, fast generation, significant heterogeneity and complexity. Therefore, more effective, flexible, and efficient clustering approaches are required. Recently, the parallel and distributed computing concepts gives birth to the parallel clustering algorithms. Nowadays, the researches focus on scalable clustering methods based on different acceleration platforms to deal with big data problems. The acceleration platforms can be classified into horizontal and vertical-scaling platforms. In this paper we present a recent overview of the latest parallel and distributed clustering algorithms based on vertical scaling platforms. Otherwise, the paper gives a discussion that will be useful for researchers to propose more effective and efficient algorithms for Big Data clustering.