AbstractsComputer Science

Certain investigations on data Clustering using hybrid algorithms for Unlabeled data sets;

by Komarasamy G

Institution: Anna University
Department: Certain investigations on data Clustering using hybrid algorithms for Unlabeled data sets
Year: 2015
Keywords: Data mining; K means; K Means Particle Swarm Optimization; Particle Swarm Optimization
Record ID: 1184279
Full text PDF: http://shodhganga.inflibnet.ac.in/handle/10603/37796


Data mining is a process of extracting knowledge from homogeneous newlinewide variety of datasets It is mainly used in interdisciplinary subfield namely newlineartificial intelligence machine learning statistics and database systems of newlinecomputer science for discovering original patterns Clustering is one of the newlineessential process of data mining The cluster analysis or clustering is the process of newlinecombining a set of items into same group and their relationships The K means newline KM algorithm is a major role in determine the number of clusters k for large newlineDatasets It needs to predefine the k value itself which is difficult and it is hard to newlinecalculate before the number of clusters that would be there in data There are no newlinecompetent and universal methods to select the best number of clusters the value newlineselected as random The key challenge in the clustering process is sensitive to the newlineselection of the initial partition in order to overcome this issue implement the newlinehybrid algorithms to select best number of clusters newlineThe Particle Swarm Optimization PSO algorithm successfully newlineconverges during the global search initial stages but around global optimum the newlinesearch process will become very slow The KM algorithm can achieve faster newlineconvergence to get the optimum solution The K Means Particle Swarm newlineOptimization KMPSO algorithm newline newline%%%reference p165-172.