Evaluation performance of time series approach for forecasting air pollution index in Johor, Malaysia

Abstract

The air pollution index (API) has been recognized as one of the important air quality indicators used to record the correlation between air pollution and human health. The API information can help government agencies, policy makers and individuals to prepare precautionary measures in order to eliminate the impact of air pollution episodes. This study aimed to verify the monthly API trends at three different stations in Malaysia; industrial, residential and sub-urban areas. The data collected between the year 2000 and 2009 was analyzed based on time series forecasting. Both classical and modern methods namely seasonal autoregressive integrated moving average (SARIMA) and fuzzy time series (FTS) were employed. The model developed was scrutinized by means of statistical performance of root mean square error (RMSE). The results showed a good performance of SARIMA in two urban stations with 16% and 19.6% which was more satisfactory compared to FTS; however, FTS performed better in suburban station with 25.9% which was more pleasing compared to SARIMA methods. This result proved that classical method is compatible with the advanced forecasting techniques in providing better forecasting accuracy. Both classical and modern methods have the ability to investigate and forecast the API trends in which can be considered as an effective decision-making process in air quality policy.