Explanation of Air Pollution Using External Data Sources

During the last years, high emission of fine-grained particles into the atmosphere and its negative impact on people’s health and well-being has attracted the attention of researchers and governmental agencies to look for the causes of air pollution in different neighbourhoods [7]. Serious measures have been taken in order to sustain the levels of air pollution, such as the introduction of fine-grained particle concentration thresholds or driving bans for vehicles that use diesel engines in several European cities [8]. When it comes to current approaches on predictive modeling in the area of air pollution, many focus on estimating the concentration of fine particulate matter in the nearest future in a particular area [2]. However, identifying the cause of high emission of fine particulate matter, as well as finding its potential sources can provide decision makers with valuable information for the design of counter measures. Detecting the sources of air pollution and treating them is a big step toward better air quality [3]. The problem we observe is that historical records from air quality sensors that are used to forecast the concentration of fine particulate matter are not sufficient for inference of factors that are likely to cause air pollution. Intuitively, we can assume that traffic, factories and production facilities, agriculture etc. might negatively affect the air quality. To test these assumptions, we need to incorporate external data sources into the main dataset of air quality sensory readings (Section 2). For this project, we aim at designing a proto-