# Rigorous Performance Analysis of State-of-the-art TSP Heuristic Solvers The dataset contains all symmetric TSP instances, performance results and features sets for "Rigorous Performance Analysis of State-of-the-art TSP Heuristic Solvers", P. McMenemy, N. Veerapen, J. Adair, G. Ochoa. The 18th European Conference on Evolutionary Computation in Combinatorial Optimisation (EvoCOP 2018), 24 - 26 April 2019, Leipzig, Germany. The dataset contains tsp format files for all 1893 instances used in the study, as well as the performance data for each TSP instance for the 3 TSP heuristics analysed. The full list of files is: Instances Files: MORPH.zip, NETGEN.zip, RUE.zip, TSPLIB.zip Performance Files: EAX_PERFORMANCE.zip, LKH_PERFORMANCE.zip, GPX2_PERFORMANCE.zip Features Files: PIHERAMUSLIU_FEATURES.zip, TSPMETA_FEATURES.zip # How to Interpret the Data The files MORPH.zip, NETGEN.zip, RUE.zip, TSPLIB.zip are of .tsp standard format. Each file is comprised of a header and a full list of "city" location coordinates. For example, one of the MORPH instances has header: NAME : 500-2-1.tsp---to---500-102.tsp COMMENT : netgen, N=500, CL=1, seed=NA TYPE : TSP DIMENSION : 500 EDGE_WEIGHT_TYPE : EUC_2D The NAME of NETGEN TSP files is of the form "size-number_of_clusters_seed.tsp", while the MORPH types are of the form "NETGEN_instance---RUE_instance.tsp" where "NETGEN_instance" and "RUE_instance" are the .tsp instances used to generate the particular MORPH instance. COMMENT describes the TSP instance defined by the NODE_COORD_SECTION below the header. TYPE is consistently "TSP". DIMENSION states the size of the TSP instance, i.e., the number of "cities" in the instance. EDGE_WEIGHT_TYPE describes the metric used to define the distance from one city to aonther. TSPLIB TSP files are as obtained from online sources at: http://www.math.uwaterloo.ca/tsp/vlsi/index.html, http://elib.zib.de/pub/mp-testdata/tsp/tsplib/tsp/, http://www.math.uwaterloo.ca/tsp/world/countries.html . The performance files EAX_PERFORMANCE.zip, LKH_PERFORMANCE.zip, GPX2_PERFORMANCE.zip contain the 30 runs of each pertinent heuristic carried out for all instances in the study. The files are in .RData format for use with the open source software, R. The files containg the extracted features sets, PIHERAMUSLIU_FEATURES.zip and TSPMETA_FEATURES.zip are in two different formats. PIHERAMUSLIU_FEATURES.zip is in comma separated values (.csv) format, while TSPMETA is of .RData format.