Adaptive black widow optimisation algorithm for data clustering

Generally, local search approaches are better than population-based approaches in exploiting the search space and worse for exploration. Recently, the black widow optimisation (BWO) algorithm was proposed for engineering optimisation problems as an algorithm balancing between the exploration and exploitation phases. However, the BWO algorithm employed three essential parameter rates, in which a different experiment is needed for each problem. The adaptive black widow optimisation (ABWO) algorithm is proposed to tune the parameters adaptively and accept worse solutions by relying on a local search, using the solutions' qualities average. Six medical datasets are used as a test domain with two calculations criteria to calculate the minimal distance. In order to evaluate the effectiveness of ABWO, a comparison is made between ABWO, BWO and other methods' performances that have been drawn from the acknowledged literature. Outcomes show ABWO is capable of obtaining better cluster qualities, thus outperforming many other methods.