Feature selection for air quality forecasting: a genetic algorithm approach

摘要: Feature selection is a process of determining the most relevant features given problem in order to improve generalization and performance classification or regression algorithm.This paper focuses on exploitation genetic algorithm following wrapping iterative approach used extract an optimal feature subset large database containing pollutant concentration measurements. The fed machine learning predict daily maximum two air pollutants.The encoding complexity representation genomes tackled. Results experimentation specific dataset quality forecasting are presented, as well some proposed alterations standard that guided mature convergence gave good solutions for this problem. A modified version initial presented well, implemented purpose being compared equal basis with other methods. Two such methods filtering type, CFS ReliefF, with.The comparative results suggest type technique described significantly better at hand, but conclusion limited uses its core phase.