Abstract
The air pollution problem which has arisen in developed countries, becoming evident by high levels of smoke from industries or traffic, has forced authorities to search for mechanisms to control the air quality by the real-time monitoring system. For this purpose, in this paper we develop a new procedure able to analyze this real-time data. More precisely, we use the recent development on mathematical Statistics to analyze the relationship between the maximum ozone concentration and the other palling gases such as the Nitric Oxides (NO), Nitrogen Dioxide (NO2) and Sulphur Dioxide (SO2). Specifically, we propose three models which are, Functional Nonparametric Regression, Functional Robust Regression and Functional Relative Error Regression. Considering the daily- curve of the concentration of the previous gases collected by the Marylebone road monitoring site in London, we provide statistical models allowing the prediction of the maximum ozone concentration 4 h ahead. We show that the accuracy of our prediction approaches is closely linked to the choice of the regression model and the input variables or the covariates. In particular, the nonparametric regression is more performant than the other models when the regressors are NO2 and SO2.