A comprehensive evaluation of air pollution prediction improvement by a machine learning method

Modeling PM2.5 Urban Pollution Using Machine Learning and Selected Meteorological Parameters

“…The application of numerical models for complex terrain regions is challenging, since important topographic features are not well represented [11,33]. This produces imprecisions in not only forecasting air quality, but also relevant meteorology [10,12,34,35]. Here, the proposed model provides a more reliable and more economical alternative to predict PM 2.5 levels, as it only requires meteorological data acquisition.…”

“…Chemical transport and Atmospheric Dispersion Modeling are numerical methods, and the most advanced ones are WRF-Chem and CMAQ. These models can be used to predict atmospheric pollution, but their accuracy relies on an updated source list that is very difficult to produce [10]. In addition, complex geophysical characteristics of locations with complex terrain complicate the implementation of these models of weather and pollution forecast mostly due to the complexity of the air flows (wind speed and direction) around the topographic features [11,12].…”

“…The most popular classifiers to forecast pollution from meteorological data are artificial Neural Networks [13][14][15]. Other successful studies use hybrid or mixed models that combine several artificial intelligence algorithms, such as fuzzy logic and Neural Network [16], or Principal Component Analysis and Support Vector Machine [17], or numerical methods and machine learning [10].…”

“…Recent studies show that the machine learning approach seems to overcome the other two methods for forecasting pollution [9,10]. This is the reason why it is increasingly used to predict air quality [13,[17][18][19][20][21].…”

“…Since machine learning is a very promising method to forecast pollution, we propose applying this approach to predict PM 2.5 concentration in Quito. This prediction is based on a selection of meteorological features for two main reasons: first because a model using only meteorological data, which can be easily obtained in any urban area, is cheaper than an air quality monitoring system and second because a general model that may work for any city is not realistic [10], which implies that a selection of meteorological parameters must be performed in order to find the best model for the capital city of Ecuador. Quito is located in the Andes cordillera in the tropical climate zone, characterized by two seasons with different accumulation of precipitation.…”