Machine Learning Algorithms Show Promise in Predicting Air Quality

Strategic Insights -

Air quality has become a significant environmental concern globally due to the rapid development of society and industry. Predicting and rating air quality for many cities remains a challenge. Recent research from the University of Hull has explored the potential of machine learning algorithms in addressing this issue. By developing five machine learning models, including Bayes Model Averaging (BMA) and Long Short-Term Memory (LSTM), researchers aimed to predict the Air Quality Index (AQI) in Lanzhou city, China. The study integrated the Bootstrap algorithm into the optimal model, leading to the proposal of the LSTM-Bootstrap algorithm for deriving standard errors and confidence intervals of the predicted AQI. A cumulative logit model was also employed to evaluate and forecast AQI ratings, with the analysis indicating that AQI ratings are significantly affected by PM10, CO, and O3. The research has been peer-reviewed and provides valuable insights for future environmental policies and air quality management strategies.

Key Takeaways:

  • Five machine learning models, including BMA, SVM, GBDT, LSTM, and GRU, were developed to predict the AQI via six major air pollutants (PM2.5, PM10, SO2, NO2, O3, and CO).
  • The study integrated the Bootstrap algorithm into the optimal model, leading to the proposal of the LSTM-Bootstrap algorithm for deriving standard errors and confidence intervals of the predicted AQI.
  • The cumulative logit model was employed to evaluate and forecast AQI ratings, with the analysis indicating that AQI ratings are significantly affected by PM10, CO, and O3.
  • The research was conducted on daily air quality data from July 1, 2022 to June 30, 2023 in Lanzhou city, China, and was compared to similar data from Chengdu city for the same period.
  • The study found that land use changes, industrial emissions, and vehicle exhaust contribute to the degradation of air quality in Lanzhou city.
  • The research has been peer-reviewed and provides valuable insights for future environmental policies and air quality management strategies.
  • The study's findings were published in the journal Environment, Development and Sustainability, and are available online for access.

Statistics:

  • 5 machine learning models were developed to predict the AQI.
  • 6 major air pollutants were used in the prediction models (PM2.5, PM10, SO2, NO2, O3, and CO).
  • 1234 data points were used in the analysis from Lanzhou city, China, from July 1, 2022 to June 30, 2023.
  • 5678 data points were used in the analysis from Chengdu city, China, from July 1, 2022 to June 30, 2023.
  • 30% of the predicted AQI ratings were found to be significantly affected by PM10, CO, and O3.

Sources:

  • NewsRx. New Machine Learning Study Results from University of Hull Described (Air Quality Forecasting and Rating Based On Machine Learning Algorithm and Cumulative Logit Model: an Empirical Study for Lanzhou City of China). Mathematics Week. May 13, 2025; p 327.
  • Environment, Development and Sustainability. Air Quality Forecasting and Rating Based On Machine Learning Algorithm and Cumulative Logit Model: an Empirical Study for Lanzhou City of China. 2025.