Scholarlink

<p> <?tk 1?>The number of sunspots is an important indicator of solar activity, which has an impact on space weather and the Earth’s climate. Hence, sunspot number forecasting is an integral part of solar cycle monitoring for the National Aeronautics and Space Administration, the European Space Agency and other space and environmental agencies. The advent of novel machine learning tools creates new opportunities for modeling and prediction of time series datasets. This paper compares predictions of the yearly sunspot number time series with five different machine learning methods in terms of prediction performance, computational speed, and interpretability. The methods explored involve a classical time series approach (Seasonal autoregressive integrated moving average), a stochastic approach (Gaussian process regression), a method based on decision trees (Light gradient boosting machine), and a long-short-term memory neural network. The relative strengths and weaknesses of each method are discussed.</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

Prediction of yearly mean sunspot number using machine learning methods

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年6月7日

查看全文
  • Nikolaos Paraskakis,
  • Dionissios T. Hristopulos

摘要

The number of sunspots is an important indicator of solar activity, which has an impact on space weather and the Earth’s climate. Hence, sunspot number forecasting is an integral part of solar cycle monitoring for the National Aeronautics and Space Administration, the European Space Agency and other space and environmental agencies. The advent of novel machine learning tools creates new opportunities for modeling and prediction of time series datasets. This paper compares predictions of the yearly sunspot number time series with five different machine learning methods in terms of prediction performance, computational speed, and interpretability. The methods explored involve a classical time series approach (Seasonal autoregressive integrated moving average), a stochastic approach (Gaussian process regression), a method based on decision trees (Light gradient boosting machine), and a long-short-term memory neural network. The relative strengths and weaknesses of each method are discussed.