Data for EMO2023 Paper "Feature-based Benchmarking of Distance-based Multi/Many-objective Optimisation Problems: A Machine Learning Perspective"

<strong>Data for Paper "Feature-based Benchmarking of Distance-based Multi/Many-objective Optimisation Problems: A Machine Learning Perspective"</strong> <br> The file <strong>dbmopp_dataset_perf.csv</strong> contains results from the 945 x 30 instances, with the following columns: <em>design_id</em>: problem identifier <em>n_var</em>: number of variables {2, ..., 20} <em>n_obj</em>: number of objectives {2, ..., 10} <em>nonident_ps</em>: non-identical Pareto sets {0 (no), 1 (yes)} <em>var_density</em>: varying density {0 (no), 1 (yes)} <em>n_discon_ps</em>: number of disconnected Pareto sets {0, ..., 6} <em>n_local_fronts</em>: number of local fronts {0, ..., 6} <em>n_resist_regions</em>: number of dominance resistance regions {0, ..., 6} <em>instance_id</em>: instance (fold) identifier {1, ..., 30} <em>budget</em>: number of evaluations performed by the algorithm {5000, 10000, 30000, 50000} <em>algo</em>: multi-objective evolutionary algorithm {NSGAII, IBEA, MOEAD, Random} <em>hypervolume</em>: hypervolume reached by the algorithm [0.0, 1.0] The file <strong>dbmopp_dataset_perf_aggregated.csv</strong> contains average results from the 945 problems, with the following columns: <em>design_id</em>: problem identifier <em>n_var</em>: number of variables {2, ..., 20} <em>n_obj</em>: number of objectives {2, ..., 10} <em>nonident_ps</em>: non-identical Pareto sets {0 (no), 1 (yes)} <em>var_density</em>: varying density {0 (no), 1 (yes)} <em>n_discon_ps</em>: number of disconnected Pareto sets {0, ..., 6} <em>n_local_fronts</em>: number of local fronts {0, ..., 6} <em>n_resist_regions</em>: number of dominance resistance regions {0, ..., 6} <em>budget</em>: number of evaluations performed by the algorithm {5000, 10000, 30000, 50000} <em>algo</em>: multi-objective evolutionary algorithm {NSGAII, IBEA, MOEAD, Random} <em>hypervolume_avg</em>: average hypervolume reached by the algorithm [0.0, 1.0] <em>best</em>: 1 if the corresponding algorithm obtains the best average hypervolume, 0 otherwise