Rules Discovery of High Ozone in Klang Areas using Data Mining Approach

Abstract

Ground level ozone (O3) is one of the common pollution issues that has a negative influence on human health. However, the increasing trends in O3 level nowadays which due to rapid development has become a great concern over the world. Thus, developing an accurate O3 forecasting model is necessary. However, the interesting pattern from the data should be identified beforehand. Association rules is a data mining technique that has an advantage to discover frequent patterns in a dataset, which subsequently will be useful in the research domain. Therefore, this paper presents the discovering knowledge based on association rules and clustering technique towards a climatological O3 dataset. In this study, the data was analysed to find the behaviour of each precursors. Later K-means clustering technique was used to find the suitable range for each chosen variable independently, then applied Apriori based association rules technique to present the behaviours in a meaningful and understandable format. The climatological O3 time series data has been collected from Department of Environment for Klang station from year 1997 to 2012. However, the proposed method only applied on high O3 concentration data during stated years to find the association pattern. The outcome has discovered 17 strong rules. The patterns and behaviours of the selected variables during high O3 concentration has been discovered. The rules are benefit to the government on how to control the air quality later.