Local neighbour spider monkey optimization algorithm for data clustering

Abstract Data clustering plays a crucial role in the analysis of information collected from a variety of domains. Researchers developed many classical and mathematical algorithms to solve real-life problems, but due to the inherent property of these algorithms, they prematurely converge and fall to...
Ausführliche Beschreibung

Abstract Data clustering plays a crucial role in the analysis of information collected from a variety of domains. Researchers developed many classical and mathematical algorithms to solve real-life problems, but due to the inherent property of these algorithms, they prematurely converge and fall to local optima. A further pattern of data in terms of shape, size, and distribution has a significant effect on the exploitation and exploration characteristic of algorithms which draw attention to many researchers. This work attempts to solve this problem by proposing an LNSMO local neighbour spider monkey optimization algorithm for data clustering. In the proposed algorithm Local Leader Phase of the spider monkey optimization algorithm is improved with its neighbour solution. Further to enhance the global search global leader phase of spider monkey optimization is improved with a chaotic operator. The performance of LNSMO is compared with eleven real-life datasets with five well-known Meta-heuristic algorithms in terms of a sum of within-cluster distance and convergence speed. It is further compared with recently developed hybrid meta-heuristic algorithms. Experimental result demonstrates that the proposed algorithm provides a better result in terms of Accuracy, F-measure, and SWCD. Ausführliche Beschreibung