Hybrid seasonal ARIMA and artificial neural network in forecasting Southeast Asia city air pollutant index

Abstract

The rise of air pollution has received much attention globally. As an early warning system for air quality control and management, it is important to provide precise future concentrations pollutant information. Using time series forecasting methods, the forecast of daily Air Pollutant Index (API) is presented here. The hybrid method between seasonal autoregressive integrated moving average (SARIMA) and artificial neural network (ANN) are chosen. To verify, the accuracies are measured using error magnitude approach. However, evaluation of forecasting API is also in fluenced by the health classification based on the threshold value assigned in air quality guidelines. Thus, forecast accuracies based on index value, namely as true predicted rate (TPR), false positive rate (FPR), false alarm rate (FAR) and successful index (SI) are also used for forecast validation. As shown in the results, the hybrid model performs better in both model's evaluations group used. Hence, the hybrid method must be considered in the forecasting area due to the capability to analyze real data consisting of both linear and nonlinear patterns. Besides, using the appropriate measurement in accordance to the purpose of forecasting is important to produce an accurate forecast.