Air quality prediction in Mexico City

Description of the service

In large cities, people's health and concentrations of particles smaller than 10 and 2.5 micrometers (PM10, PM2.5) and ozone (O_3) are correlated, so their prediction is useful for the government and citizens. Mexico City has an air quality prediction system with a forecast by pollutant per time zone and geographic area, but it is only operative for the next 24 hours.

Predictions for longer periods rely on sophisticated methods, but highly automated techniques, such as deep learning require a large amount of data, not available for this case. Therefore, a set of predictor variables feeds and tests different Machine Learning (ML) methods and determines the essential features for predicting different pollutant concentrations to develop an ad-hoc hybrid model that includes ML features but allows a level of explainability, unlike what would occur with methods such as neural networks.

Different statistical methods and ML techniques allowed the creation of an initial model to estimate the concentration of the three major pollutants in the air of Mexico City two weeks in advance.