Proposal of a Heuristic for Cluster Analysis with Application in Allocation of Anaerobic Co-digesters for Biogas Production

Cluster analysis refers to optimally segmenting a set of entities, and there are several types of algorithms to reach this purpose. It can be applied in decision making, machine learning, data mining, and pattern recognition. This work aims to develop a simple and fast partition heuristic to solve the cluster analysis problem, where the optimality criteria are the smallest distances between the vertices of the same cluster. To verify the proposed heuristic, tests were made in instances ranging from 20 to 5000 nodes and different degrees of complexity. The developed algorithm can be applied to large size datasets were time could be a limitation factor. As an example of application, the heuristic developed was used to choose the location for installing anaerobic co-digesters for the production of biogas in the state of Paraná, Brazil.