State-space Time Series Analysis on Air Pollution Data

Abstract

Nowadays, statistical modeling of air pollution data is an important topic, particularly for the purposes of forecasting and risk assessment. Thus, this study proposes the application of a univariate state space model in analyzing the time series data of air pollution. Several useful functions and packages available in R software for an easy application of the state space model are discussed. In a similar vein, several illustrative examples covering fitted local-level and local linear trend models, which particularly use the StructTS function, are also presented. A case study is conducted using the data of air pollution index (API) in Klang, Malaysia. Based on the model comparison and diagnostic evaluation, the results find that a local-level model is sufficient in providing a good fitted model to describe the behaviors of API data in Klang. However, in order to provide a better evaluation, we suggest that the state space model must be re-estimated to obtain the latest forecasting assessment of API values over time. To conclude, the state space model may be used as a good alternative tool for air quality forecasting.