Improved Air Pollution Forecasting With Hybrid Machine Learning Algorithms

Abstract

The prediction of accurate pollutant in any dynamic nature provides different newlinebenefits for humans or other living creatures on the planet. The machine learning process newlineand classification technique are developed for forecasting the air pollution that attains newlineenhanced result on cloud computing environment. The accurate air pollutant forecasting newlineperformance of data is an important issue which affects the air quality. During the newlineprediction of air pollution, feature selection and data classification is the most considerable newlinetask for accurate forecasting results. The performance of feature selection helps to give newlineenhanced results in prediction with higher accuracy. Here, irrelevant features are newlineeliminated effectively. By removing negative features of air data, people s health newlineconditions are managed to prevent the dangerous impact. newlineSeveral machine learning methods are performed on air data for estimating newlinepollution forecasting performance. But the accurate forecasting with minimum time and newlineminimum memory consumption is a challenging issue. The accurate performance of newlinepollution prediction is an important goal in an environmental issue. Accurate pollution newlineforecasting is a demanding problem to predict air quality index at particulate levels. newlineHence, three different models, namely, Linear Regression and Multiclass Support Vector newline(LR-MSV) model 1, Bilateral Transformative Broken Stick Regression and Quadratic newlineWeighted Emphasis Boost Classification (BTBSR-QWEBC) model 2 and Discretized newlineRegression and Least Square Support Vector (DR-LSSV) model 3 are developed with newlinevarious air quality data from dataset to attain efficient air pollution forecasting for newlinemonitoring and controlling the air pollution. newlineThe LR-MSV model is proposed at first for forecasting air pollution with accurate newlineresults. The performance of pre-processing is handled on each input data through sliding newlinewindow to remove noise data for reducing time of classification. Subsequently, LRC newline(LRC) is utilized to select the relevant features of data. Following this, the corre