Scholarlink

<p>Industrial facilities are obligated to report their pollutant emissions, wastewater discharges, and waste generation to the European Pollutant Release and Transfer Register (E-PRTR). The German PRTR register has demonstrated a commitment to transparency regarding pollution data, as evidenced by the receipt of over 80,000 notifications from more than 5000 companies between 2007 and 2022. This figure, along with the substantial volume of data reported, positions the register as a valuable source of information for researchers and stakeholders interested in environmental issues. This study explores the viability of the PRTR dataset as a data source for automated predictions of future pollutant releases using machine learning (ML). In accordance with the CRISP-DM process model, an analytical dataset is constructed and utilized to train two machine learning models: one to predict the annual branch-specific top 10 polluters for greenhouse gas (GHG) emissions into the air, and another for heavy metal discharges into water. The findings indicate that both models generate predictions that are closely aligned with the actual pollution reports. These predictions have the potential to provide valuable insights and support to a range of stakeholders, including corporate environmental managers, regulatory agencies, business organizations, and authorities engaged in combating environmental crimes.</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

Predictive modeling of major polluters based on the German PRTR emissions data

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年7月15日

查看全文
  • Heiko Thimm,
  • Vanessa Schmidt

摘要

Industrial facilities are obligated to report their pollutant emissions, wastewater discharges, and waste generation to the European Pollutant Release and Transfer Register (E-PRTR). The German PRTR register has demonstrated a commitment to transparency regarding pollution data, as evidenced by the receipt of over 80,000 notifications from more than 5000 companies between 2007 and 2022. This figure, along with the substantial volume of data reported, positions the register as a valuable source of information for researchers and stakeholders interested in environmental issues. This study explores the viability of the PRTR dataset as a data source for automated predictions of future pollutant releases using machine learning (ML). In accordance with the CRISP-DM process model, an analytical dataset is constructed and utilized to train two machine learning models: one to predict the annual branch-specific top 10 polluters for greenhouse gas (GHG) emissions into the air, and another for heavy metal discharges into water. The findings indicate that both models generate predictions that are closely aligned with the actual pollution reports. These predictions have the potential to provide valuable insights and support to a range of stakeholders, including corporate environmental managers, regulatory agencies, business organizations, and authorities engaged in combating environmental crimes.