Air Pollution Forecasting using Novel Feature Selection and Classification Model

Abstract

Quality of air prediction was the difficult function towards dynamic nature, instability, as well as better inconsistency within the time and space. Especially in urban areas, Air Pollution is becoming more and more, and the various pollutants affect the air quality. Hence, it is essential to accurately predict Air Pollution for providing hazardous impact earlier. The existing machine learning methods have been developed but it is difficult to forecast accurate pollutant and particulate levels and to predict the air quality index. Bilateral Transformative Broken-Stick Regression-based Quadratic Weighted Emphasis Boost Classification (BTBSR-QWEBC) technique is introduced for IoT-based Air Pollution Forecast with higher accuracy and minimum time consumption for increasing accuracy of air pollution forecasting. From the BTBSR-QWEBC, IoT devices are used to collect Air Quality data. The BTBSR-QWEBC technique includes three major processes namely pre-processing, Feature Selection, and classification. That Technique helps to improve the accuracy of the Air Pollution Forecast and to minimize time consumption. Experimental assessment is performed by various metrics namely Air Pollution Forecast accuracy, error rate, as well as Air Pollution Forecasting time and space complexity. The observed results display the BTBSR-QWEBC technique provides better accuracy as well as minimal time than conventional techniques.