An efficient hybrid data clustering method based on Candidate Group Search and genetic algorithm

Data Mining is an efficient data analysis process which is used to find the patterns and relationship of a large database. Clustering is a popular technique of data mining for unsupervised learning in which labels are not defined previously. K-Mean is a well known partitioning technique for forming different clusters, but it has the drawback of initial sensitivity and local optima convergence. K-Harmonic algorithm solves the initial sensitivity problem, but it stuck in local optima problem. Genetic algorithm is an efficient tool of the search and optimization problems, which offers the benefits like selective search. In this paper, presents a new scheme in which the initial centroids are calculated using the Candidate Group Search which results in reduction of time for genetic process. Genetic algorithm is used to assign the data elements to the suitable cluster. The experimental results showed that the proposed scheme automatically finds the cluster centers and reaches to global optimal solution for data clustering.