Scholarlink

<p>Air contaminants from pollution are still an increasingly significant climatological and public well-being issue, with small particulate matter (PM2.5) majorly adding to harmfulness and death rates across the globe. In the past few decades, computer-based machine learning has surfaced as a crucial instrument in forecasting PM2.5 quantity, leading to preemptive regulations for contaminant containment. This review’s objective is to evaluate the efficiency of multiple machine learning networks, Support Vector Machines, Random Forests, and Artificial Neural Networks in forecasting PM2.5 amounts. The research analyses the architecture of each model as well as the functioning inputs and dynamic implications to decide each one's comparative power and fragility. The evaluation encompasses a comprehensive differentiation of network abilities spanning many areas and information outlines, displaying combined networks like artificial neural network autoregressive integrated moving average (ANN-ARIMA), Support vector machine genetic algorithm (SVM-GA), and Random forest long short term memory (RF-LSTM). This analysis also exemplifies procedures utilised in information compounding, input selection, and chrono-spatial examinations. Various research from areas like China, the U.S., Europe, and India gives room for model precision and malleability. Also, the examination traverses provocations such as information sparsity, network relevance, and interpretability, providing information for the future. Finally, this review ends up highlighting the fact that combination and crossbred machine learning models provide far better precision in PM2.5 prediction. Folding in dynamic information and better-improved methods greatly influence model trustworthiness. From the authors’ standpoint, encompassing machine learning into climatological regulations in terms of government is an effective route to the well-being of public health.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A general review on the applications of machine learning to PM2.5 air pollution forecasting

  • 期刊论文

  • 发表日期：

    2025年9月3日

查看全文
  • Swara Patel,
  • Manan Shah,
  • Krishaan Patel,
  • Mitul Prajapati

摘要

Air contaminants from pollution are still an increasingly significant climatological and public well-being issue, with small particulate matter (PM2.5) majorly adding to harmfulness and death rates across the globe. In the past few decades, computer-based machine learning has surfaced as a crucial instrument in forecasting PM2.5 quantity, leading to preemptive regulations for contaminant containment. This review’s objective is to evaluate the efficiency of multiple machine learning networks, Support Vector Machines, Random Forests, and Artificial Neural Networks in forecasting PM2.5 amounts. The research analyses the architecture of each model as well as the functioning inputs and dynamic implications to decide each one's comparative power and fragility. The evaluation encompasses a comprehensive differentiation of network abilities spanning many areas and information outlines, displaying combined networks like artificial neural network autoregressive integrated moving average (ANN-ARIMA), Support vector machine genetic algorithm (SVM-GA), and Random forest long short term memory (RF-LSTM). This analysis also exemplifies procedures utilised in information compounding, input selection, and chrono-spatial examinations. Various research from areas like China, the U.S., Europe, and India gives room for model precision and malleability. Also, the examination traverses provocations such as information sparsity, network relevance, and interpretability, providing information for the future. Finally, this review ends up highlighting the fact that combination and crossbred machine learning models provide far better precision in PM2.5 prediction. Folding in dynamic information and better-improved methods greatly influence model trustworthiness. From the authors’ standpoint, encompassing machine learning into climatological regulations in terms of government is an effective route to the well-being of public health.