Hierarchical clustering-based framework for a posteriori exploration of Pareto fronts: application on the bi-objective next release problem

• DOI: 10.3389/fcomp.2023.1179059
• Publication Date: 2023-05-24T04:51:00Z
• Authors: Carlos Casanova; Esteban Schab; Lucas Prado; Giovanni Daián Ro...
• Keywords: Benchmark (surveying); Hierarchical clustering; Dashboard

Introduction When solving multi-objective combinatorial optimization problems using a search algorithm without priori information, the result is Pareto front. Selecting solution from it laborious task if number of solutions to be analyzed large. This would benefit systematic approach that facilitates analysis, comparison and selection or group based on preferences decision makers. In last decade, research development algorithms for has been growing steadily. contrast, efforts in posteriori exploration non-dominated are still scarce. Methods paper proposes an abstract framework hierarchical clustering order facilitate makers explore such front according their preferences. An extension aimed at addressing bi-objective Next Release Problem presented, together with Dashboard implements extension. Based this implementation, two studies conducted. The first usability study performed small experts. second performance analysis computation time consumed by algorithm. Results results initial empirical promising indicate directions future improvements. experts were able correctly use dashboard properly interpret visualizations very short time. same direction, highlight advantage clustering-based terms response Discussion these excellent results, new planned, as well implementation validity tests expert real-world data.