Cooperative clustering algorithm based on brain storm optimization and K-means
Eva Tuba, Ivana Strumberger, Nebojša Bačanin, Dejan Živković, Milan Tuba · 2018
Data analysis and making prediction models are important tasks in numerous fields such as medicine, economy, marketing and others. Data clustering provides useful information and it is a rather common tool for discovering data properties. K-means is one of the simplest clustering algorithm but its severe flow is getting trapped into local optima, hence it can be improved by introducing global search. In this paper, cooperative algorithm based on the brain storm optimization algorithm and k-means is proposed. Local search in brain storm optimization algorithm used for solving clustering problems is improved by introducing one iteration of the k-means algorithm for each generated solution. The proposed method was compared with five nature inspired clustering algorithms and by the basic brain storm optimization. Brain storm optimization combined with k-means algorithm found better solutions, smaller fitness function values, and also reduced execution time compared to other methods from literature.