Time series motif and pattern mining to analyse air pollution data

Abstract

Air pollution is a global issue in our contemporary civilization, and it is the primary cause of impending global warming. The main cause of air pollution is the growing economic condition and the high number of chemical industries in urban and suburban areas. Although a rise in the number of automobiles and changing lifestyle are also major contributors to air pollution. Respiratory disease, heart disease, irritation of the eyes, breathlessness, etc. are some of the dangerous effects of air pollution. Because of the significant growth of air pollutants, government agencies have set up pollution monitoring stations throughout the country using sensors, and those sensors generate a large amount of data and that data may be visualized as time series data for pollution pattern identification to better understand variability in pollution levels. In the above-mentioned context, the focus of this project is to collect real world air pollution data from government setup pollution monitoring stations of Delhi and find temporal motif (i.e., pattern) in the pollutants’ time series. Specifically, our study focuses on identifying consensus motif for various subsequence lengths. An existing motif discovery algorithm such as ostinato, MASS, etc. are employed to achieve the project goal.