Modified bee colony optimization with opposition learning algorithm on use of medical data clustering

Clustering has gained popularity in the data mining field as one of the primary approaches for obtaining data distribution and data analysis. The medical data analysis for different diseases is a great challenge in current research. The benefits of opposition based learning such as faster convergence rate and better approximate result in finding global optimum can be helpful in this area. To achieve faster convergence and better clustering results for medical data, in this work, the authors have proposed an approach utilising modified bee colony optimization with opposition based learning and k-medoids technique. The initial centroid plays an important role in the bee colony optimization based clustering. The proposed approach uses k-medoids algorithm for this task. In order to facilitate faster convergence, it adds the opposite bees which are located at exactly the opposite location of the initial bees. The exploration task is performed by both of these kinds of bees to find potential solutions. This increases the algorithm's capacity for exploration and, consequently, the rate of convergence. Five distinct medical datasets collected from the UCI library are investigated to demonstrate the algorithm's efficacy. The implementation results demonstrate that the algorithm gives better convergence rate and clustering quality compared to some the existing algorithms.