A matheuristic for large-scale capacitated clustering

Clustering addresses the problem of assigning similar objects to groups. Since size clusters is often constrained in practical clustering applications, various capacitated problems have received increasing attention. We consider here p-median (CPMP) which p are selected as cluster centers (medians) such that total distance from these medians their assigned minimized. Each object associated with a weight, and weight each must not exceed given capacity. Numerous exact heuristic solution approaches been proposed for CPMP. The state-of-the-art approach performs well instances up 5,000 but becomes computationally expensive much larger number objects. propose matheuristic new decomposition strategies can deal comprising 500,000 In computational experiment, consistently outperformed on medium- large-scale while having performance small-scale instances. As an extension, we show our be applied related problems, centered (CCCP). For several test CCCP, found best-known solutions.