Statistical Modelling For Forecasting Pm10 Concentrations In Peninsular Malaysia

Loading...
Thumbnail Image
Date
2017-11
Authors
Ng, Kar Yong
Journal Title
Journal ISSN
Volume Title
Publisher
Universiti Sains Malaysia
Abstract
This research aims to forecast the daily average PM10 concentrations in Peninsular Malaysia by using univariate modelling, i.e. time series modelling and regression modelling. In time series analysis, a typical problem in forecasting is the underestimation of the peaks. Since the series of PM10 concentrations change rapidly, this research proposed the use of wavelet-based time series model to improve the forecast accuracy, i.e. the application of discrete wavelet transform (DWT) before the time series modelling by the Box-Jenkins autoregressive integrated moving average (ARIMA) and generalized autoregressive conditional heteroscedasticity (GARCH) models. By employing DWT, the volatile PM10 series were decomposed into several subsidiary series with smaller variations, and consequently, helped in improving the total forecast accuracy substantially especially during haze periods when compared to the time series modelling without DWT.
Description
Keywords
Statistical Modelling , Forecasting Pm10 Concentrations , Peninsular Malaysia
Citation