A Machine Learning Approach for Air Quality Prediction: Model Regularization and Optimization

In this paper, we tackle air quality forecasting by using machine learning approaches to 1 predict the hourly concentration of air pollutants (e.g., Ozone, PM2.5 and Sulfur Dioxide). Machine 2 learning, as one of the most popular techniques, is able to efficiently train a model on big data by using 3 large-scale optimization algorithms. Although there exists some works applying machine learning 4 to air quality prediction, most of the prior studies are restricted to small scale data and simply train 5 standard regression models (linear or non-linear) to predict the hourly air pollution concentration. 6 In this work, we propose refined models to predict the hourly air pollution concentration based 7 on meteorological data of previous days by formulating the prediction of 24 hours as a multi-task 8 learning problem. It enables us to select a good model with different regularization techniques. We 9 propose a useful regularization by enforcing the prediction models of consecutive hours to be close 10 to each other, and compare with several typical regularizations for multi-task learning including 11 standard Frobenius norm regularization, nuclear norm regularization, `2,1 norm regularization. Our 12 experiments show the proposed formulations and regularization achieve better performance than 13 existing standard regression models and existing regularizations. 14