A Two Stage Feature Selection Method to Enhance Prediction of Daily Pm2.5 Concentration Air Pollution in Cheras

Air pollution is a serious worldwide issues that negatively affect humans, environment and animals. It is essential to predict air quality to monitor dangerous pollutants including particulate matter, carbon monoxide and ozone. PM2.5 is a very dangerous pollutant after PM10 due to the size of PM2.5 is very small which is <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{2. 5}$</tex> micrometers or les. It can easily penetrate to our lung when we breath and lead to many chronic diseases such as lung cancer. Thus, this study aims to predict PM2.5 concentrations of the next day using machine learning approaches. However, due to high correlation between independent variables in air quality dataset, we used adjcorT feature selection method to find the best feature combinations while considering the multicollinearity issues in air quality data. Hence, we will conducted two-stages feature selection which is adjcorT-RBFNN to enhance PM2.5 concentrations predictions of the next day. The analysis was conducted by using hourly secondary data obtained from Department of Environmental Malaysia (DOE) from 2018 to 2022. The dataset consists of Air Pollutant Index (API) in Cheras and other 10 independent variables. The study's finding shows that the significant features to classify PM2.5 concentrations of the next day in Cheras are PM10, PM2.5, CO, O3, wind direction, NO2, wind speed, SO2 and relative humidity. Moreover, the proposed method, adjcorT-RBFNN outperformed RBFNN method with high value of accuracy, sensitivity, F1 score and AUROC which are <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$0.8291,0.8365,0.8292$</tex>, and 0.8291 respectively. Hence, the study hopes the two-stages feature selection method, adjcorTRBFNN can help further researcher to enhance PM2.5 concentrations predictions in other urban area and bring awareness to our governments for early precautions.