Evolving Differential evolution method with random forest for prediction of Air Pollution

Abstract

The aim of this paper is to use a heterogeneous ensemble of differential evolution with random forest method for air pollution prediction. This is different from existing work (independent classifier of Bayesian network and multi-label classifier used for the estimation of air pollutants) as a method is proposing to combine state-of-the-art differential evolution strategies with random forest method instead of focusing on existing single technique. When the existing approach i.e. independent and multi-label classifiers are compared with proposed approach, it shows proposed approach leads to the performance gains. Continuous ambient air quality data of two cities Delhi and Patna from Central Pollution Control Board were publically made available, from where seven pollutants (C6H6, NO2, O3, SO2, CO, PM2.5 and PM10) dataset are collected with daily average concentration.