A comparison of Autoregressive Moving Average ARMAand Neural Network Models for sulfur dioxide forecasting at Bukit Rambai Melaka Malaysia

Hafizan Juahir, . and Sharifuddin M. Zain, . and M. Nazari Jaafar, . and M. Talib Latif, . and Zainol Mustafa, . (2003) A comparison of Autoregressive Moving Average ARMAand Neural Network Models for sulfur dioxide forecasting at Bukit Rambai Melaka Malaysia. [Proceedings Paper]

Time series ARMA and neural network models namely backpropagation models each designed to forecast future sulfur dioxide SO2 values in Sungai Rambai were compared in this work. Six months historical May-October 1996 SO2 data were obtained from the ASMA station at Bukit Rambai Inustrial Park and were used to build these models. The time series ARMA model and neural network model are able to simulate well the historical SO2 data. The simulated values of SO2 were compared with the actual values of the training data and it is found that the neural network model is marginally better in simulating SO2 values compared to the ARMA model. The ARMA model gave a correlation coefficient of 0.77062 while the ANN model gave a correlation coefficient of 0.88326 for the training data. The future values of SO2 can then be predicted from these models.

Full text available from:

Abstract

Time series ARMA and neural network models namely backpropagation models each designed to forecast future sulfur dioxide SO2 values in Sungai Rambai were compared in this work. Six months historical May-October 1996 SO2 data were obtained from the ASMA station at Bukit Rambai Inustrial Park and were used to build these models. The time series ARMA model and neural network model are able to simulate well the historical SO2 data. The simulated values of SO2 were compared with the actual values of the training data and it is found that the neural network model is marginally better in simulating SO2 values compared to the ARMA model. The ARMA model gave a correlation coefficient of 0.77062 while the ANN model gave a correlation coefficient of 0.88326 for the training data. The future values of SO2 can then be predicted from these models.

[error in script]

Item Type:	Proceedings Paper
AGROVOC Term:	SULPHUR DIOXIDE
AGROVOC Term:	FORECASTING
AGROVOC Term:	AIR POLLUTION
AGROVOC Term:	RIVERS
AGROVOC Term:	TIME SERIES ANALYSIS
AGROVOC Term:	ANALYTICAL METHODS
AGROVOC Term:	ENVIRONMENTAL IMPACT
AGROVOC Term:	HEALTH HAZARDS
AGROVOC Term:	MALAYSIA
Geographical Term:	MALAYSIA
Depositing User:	Ms. Norfaezah Khomsan
Last Modified:	24 Apr 2025 05:27
URI:	http://webagris.upm.edu.my/id/eprint/16127

Actions (login required)

View Item