In recent years, the arisen of air pollution in urban area address much attention globally. The air pollutants has emerged detrimental effects on health and living conditions. Time series forecasting is the important method nowadays with the ability to predict the future events. In this study, the forecasting is based on 10 years monthly data of Air Pollution Index (API) located in industrial and residential monitoring stations area in Malaysia. The autoregressive integrated moving average (ARIMA), fuzzy time series (FTS) and artificial neural network (ANNs) were used as the methods to forecast the API values. The performance of each method is compare using the root mean square error (RMSE). The result shows that the ANNs give the smallest forecasting error to forecast API compared to FTS and ARIMA. Therefore, the ANNs could be consider a reliable approach in early warning system to general public in understanding the air quality status that might effect their health and also in decision making processes for air quality control and management.
Principal Component Regression and Artificial Neural Network: The Prediction of Air Pollution Index (API)
