AirCast: Improving Air Pollution Forecasting Through Multi-Variable Data Alignment

TL;DR

AirCast is a multi-variable, multi-task forecasting model that improves air pollution predictions by integrating weather and chemical data, using a novel loss function and domain-informed variable selection.

Contribution

The paper introduces AirCast, a novel multi-variable, multi-task air pollution forecasting model with a frequency-weighted loss and domain-informed variable selection, enhancing prediction accuracy.

Findings

Improved accuracy in pollution forecasting compared to baseline models.

Effective handling of rare extreme pollution events with the new loss function.

Demonstrated the importance of variable selection and data alignment in model performance.

Abstract

Air pollution remains a leading global health risk, exacerbated by rapid industrialization and urbanization, contributing significantly to morbidity and mortality rates. In this paper, we introduce AirCast, a novel multi-variable air pollution forecasting model, by combining weather and air quality variables. AirCast employs a multi-task head architecture that simultaneously forecasts atmospheric conditions and pollutant concentrations, improving its understanding of how weather patterns affect air quality. Predicting extreme pollution events is challenging due to their rare occurrence in historic data, resulting in a heavy-tailed distribution of pollution levels. To address this, we propose a novel Frequency-weighted Mean Absolute Error (fMAE) loss, adapted from the class-balanced loss for regression tasks. Informed from domain knowledge, we investigate the selection of key variables known to influence pollution levels. Additionally, we align existing weather and chemical datasets across spatial and temporal dimensions. AirCast's integrated approach, combining multi-task learning, frequency weighted loss and domain informed variable selection, enables more accurate pollution forecasts. Our source code and models are made public here (https://github.com/vishalned/AirCast.git)