A Comprehensive Survey on Centroid Selection Strategies for Distributed K-means Clustering Algorithm

Abstract

Extremely large data sets often known as "Big Data" are analyzed for interesting patterns, trends, and associations, especially those relating to human behavior and interactions. Extraction of meaningful and useful information needs to be done in parallel using advanced clustering algorithms. In this paper, effort has been made to tweak in changes to the existing K-means algorithm so as to work in parallel using MapReduce paradigm. K-means due to its gradient descent nature is highly sensitive to the initial placement of the cluster centers. This random initialization of cluster centers results in empty clusters and slower convergence. In this paper, an overview of existing methods with emphasis on computational efficiency is presented. Comparison of three well known linear time complexity initialization methods has been presented here. These methods are analyzed on two different data sets. The experimental results are recorded and presented with insights on different initialization methods for practitioners.

Issues of K Means Clustering While Migrating to Map Reduce Paradigm with Big Data: A Survey

“…Poonam Ghuli et. al [12] has done a comprehensive survey on centroid selection strategies for distributed k-means clustering algorithm. The execution is divided into four modules.…”

User online behavior based on big data distributed clustering algorithm

“…The experimental results of Jin Zhou on the Hadoop distributed platform show that the SPAB-DKMC algorithm can reduce the number of iterations and improve the efficiency of the distributed K -means clustering algorithm. 15 –26…”

“…The experimental results of Jin Zhou on the Hadoop distributed platform show that the SPAB-DKMC algorithm can reduce the number of iterations and improve the efficiency of the distributed K-means clustering algorithm. [15][16][17][18][19][20][21][22][23][24][25][26] In foreign countries, Poonam Ghuli has proposed an improved density-based distributed clustering (DBDC) algorithm. The algorithm uses the data grid mapping method to map the data objects to the local spatial grid first, which improves the efficiency of local clustering.…”

“…Experimental results show that the improved DBDC algorithm is more accurate than the DBDC algorithm. 17,18 Distributed data mining technology and data mining technology based on distributed clustering have been widely used in the past decade because they deal with massive heterogeneous data sets that cannot be collected centrally. Current distributed clustering methods typically generate global models by aggregating local results obtained at each site.…”