Abstract

Forecasting of air pollutant levels is very important in environmental science research today. The rise of air pollution in both developed and developing countries has attracted much global attention. In view of current world environment quality, this pollution may affect health, ecosystem, forest species, and agriculture. In the recent years, many researchers have studied several methods and models to predict Air Pollution Index. Therefore, this paper presents methods of artificial neural network and fuzzy time series to forecast the API values in Port Klang, Malaysia. This research also employs popular measure errors which are mean squared error, mean absolute percentage error, and root mean squared error to identify the best model with the smallest error. The result shows that ANN gives the smallest forecasting error to forecast API compared to FTS. Hence, it is proven that the ANN method can be applied successfully as tools for decision making and problem-solving in forecasting research.