Abstract
Analyzing the principal causes of air pollution is of great importance in developed countries. Air quality has a potential environmental impact on the quality of life and health of humans and animals. The most developed countries control the environmental impact of air quality using a real-time monitoring system. Indeed, this monitoring system requires a sophisticated software application constructed by statistical and/or mathematical algorithms. In this work, we study a new statistical algorithm, allowing us to examine, in real time, the evolution of the ozone emission in the City of Westminster in London. Precisely, we use a high-dimensional statistical model to analyze ozone (O-3) data as a functional time series. We define the predictor using the conditional mode estimation, which is the location point that maximizes the conditional density estimation. Several big data estimators of the functional conditional density are used to carry out this predictor. In particular, we use the Nadaraya-Watson method, the k-nearest neighbors (NN) method, and the NN local linear method to estimate this functional model. We conduct this statistical analysis to predict 1-day-ahead total ozone concentrations using the daily curve ozone emission collected at the Marylebone road monitoring site in central London. The obtained results show that the proposed predictor improves the accuracy of the standard approaches based on classical regression. In addition to this real-time prediction of air quality, the proposed models are used to determine the reference intervals of the ozone concentration in this vital city of London.