Searching Parsimony Models with Genetic Algorithms
Methodology that combines feature selection, model tuning, and parsimonious model selection with Genetic Algorithms (GA) proposed in {Martinez-de-Pison} (2015) . To this objective, a novel GA selection procedure is introduced based on separate cost and complexity evaluations.
News
Version 0.9.2 (2018-05)
Keeps and shows the individual with the best validation score in the whole GA process. It can be different than the best parsimonious solution at the last generation.
Solved problems with 'suggestions'. This parameter can be set with an initial population matrix in order to continue the GA process.
The 'object' and 'iter_ini' parameters permit to continue the GA process by using a previous 'ga_parsimony' object by selecting an initial population from '[email protected]' and with GA settings of the previous GA process.
Included a new parameter 'path_name_to_save_iter'. If it is not NULL, the 'ga_parsimony' object is saved in each iteration in a file with that name.