Air Pollution Monitoring in Intelligent Cities Using Weighted Association Rule Mining

Abstract

Through the use of internet of things-based sensors in air quality monitoring stations, concentration of different pollutants and meteorological parameters can be regularly measured. In case of unusual conditions (e.g., increased levels of dangerous pollutants), a smart assessment system can produce warning so that appropriate air quality management process can be initiated. In this context, the objective of this study is to discover relationships and patterns among air pollution features and characteristics. In this case, determination of frequently observed association rules can trigger an appropriate background smart environment system when a critical situation is detected. In the experimental studies in the current project, traditional association rule mining and weighted association rule mining methods have been employed using real-world datasets collected from 21 monitoring stations in Turkey. In consequence, useful and outstanding association rules exceeding the user-defined support and confidence levels were obtained that can form basis for further research.