Multi objective hybridized firefly algorithm with group search optimization for data clustering

Data mining advances as a promising solution in exploring knowledge concealed in database and clustering is one its application. Clustering can be explained as the unconfirmed categorization of patterns into groups. It is the task of combining a set of objects into diverse subsets such that objects belonging to the similar cluster are extremely related to each other. Various objective functions are used to measure the effectiveness of the partition by examining some intrinsic property compactness of the clusters, distance measures, cluster symmetry and density. But inclusion of these objectives may not contribute to accurate classification to clusters. Multi-objective optimization (MOO) techniques are now-a-days used as an alternative technique to yield better clustering results. In the previous method, fractional cuckoo search was employed for data clustering. In this paper, we designed to improve the clustering performance by inclusion of hybridized optimization technique which will employ firefly algorithm with Group Search Optimizer (GSO). The hybridization is carried out by replacing the worst fitness values each iteration of the GSO with the updated values from firefly algorithm. Multiple objective functions are used for the calculation of the fitness and those employed fitness are Fuzzy DB-Index, Sym-Index and XB-Index. The proposed technique is implemented using MATLAB and the achievement of the clustering technique is evaluated using different indices in CVAP tool and compared with different techniques in CVAP tool.