Using finite mixtures of M-quantile regression models to handle unobserved heterogeneity in assessing the effect of meteorology and traffic on air quality

Between 2012 and 2015, the PMetro project collected space–time aerosol and gas measurements by using instruments integrated on one of the Minimetro cabins, a public conveyance of the town of Perugia (Italy). In this work, we use PMetro data to study the effect of vehicular traffic and meteorological measurements on the distribution of fine particulate matter by fitting a finite mixture of M-quantile regression models. Using this methodology, it is possible to account for heterogeneity in the data using random effects with a discrete distribution with masses and probabilities directly estimated from the data. This allows us to relax assumptions on the parametric shape of the distribution of the random effects and grants extra-flexibility. In addition, it allows us to investigate the relationship between fine particulate matter concentration and the covariates at different levels of the conditional response distribution. Empirical results show that radon concentration and vehicular traffic have the largest effect on the distribution of fine particulate matter and provide some guidelines for policy makers.