Unleashing Realistic Air Quality Forecasting: Introducing the Ready-to-Use PurpleAirSF Dataset

TL;DR

This paper introduces PurpleAirSF, a comprehensive, open-access air quality dataset with high temporal resolution and diverse coverage, aimed at facilitating research and development of accurate forecasting models.

Contribution

The paper provides the first detailed, publicly available dataset for air quality forecasting and establishes benchmark results using various models.

Findings

PurpleAirSF enables improved model validation and comparison.

Preliminary experiments establish baseline performance for future research.

The dataset covers diverse geographical regions with high temporal resolution.

Abstract

Air quality forecasting has garnered significant attention recently, with data-driven models taking center stage due to advancements in machine learning and deep learning models. However, researchers face challenges with complex data acquisition and the lack of open-sourced datasets, hindering efficient model validation. This paper introduces PurpleAirSF, a comprehensive and easily accessible dataset collected from the PurpleAir network. With its high temporal resolution, various air quality measures, and diverse geographical coverage, this dataset serves as a useful tool for researchers aiming to develop novel forecasting models, study air pollution patterns, and investigate their impacts on health and the environment. We present a detailed account of the data collection and processing methods employed to build PurpleAirSF. Furthermore, we conduct preliminary experiments using both classic and modern spatio-temporal forecasting models, thereby establishing a benchmark for future air quality forecasting tasks.