Here you will find the codes used to generate the statistics and graphs for the article:
• Leonardo Trujillo and Joel Nation, Characterizing the feature space transformations produced by genetic programming using the optimal transport dataset distance, under review in IEEE Transactions in Evolutionary Computation.
The files are separated into three folders, each one containing Matlab files (.mat) with the raw data for the classification and regression problems. These were generated with:
• GPLAB Toolbox: a Matlab toolbox based on GPLab developed by Sara Silva.
• M3GP: a framework developed by Luis Muñoz, and used in the article M3GP - Multiclass Classification with GP, code for M3GP found here, which is based on GPLAB.
• OTDD: Optimal Transport Dataset Distance, developed by David Alvarez Melis can be found here
The folders are the following:
• clas_exp: Contains the files used for classification problems
• reg_exp: Contains the files used for regression problems
• global_exp: Contains the files used for the global analysis
Each folder contains the following files with the following prefixes:
• _OTTD_m3gp (.csv): Contains all the data points of each run of the problems for clasification or regression. In the case of Global, this file contains all the runs. These .csv files contains the following information: OTDD score, percentage of improvement, three size, number of original features, M3GP generated features, PCA score and variance to mention a few.
• _stas (.py): This is the code to generate the statistical metrics (mean, median, standard deviation, quartiles, and improvement calculation).
• _p_stats (.py): This is the code to validate the statistical hypotheses (p, r, slope).
• _graphs (.py): This is the code to generate the graphs for the levels, group, problem, and run.
• gen_arbol (.py): file is also included, which is used to generate the trees for the models generated by M3GP. This file needs graphviz and scipy in order to work.
The clas_exp and reg_exp folders contain the folders with the "_mats" (.mat) prefix, which contain the raw data for the classification and regression problems created in Matlabn. To open this files in python, you will need scipy loadmat.
MIT License
Copyright (c) 2025 Joel Nation
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.