摘要
随着我国碳达峰碳中和目标的提出,碳排放审计深受关注。2022年“3·15”国际消费者权益日,国内首批碳排放数据造假案例浮出水面。我国高度重视碳排放数据造假问题,把打击数据造假、加强数据监管作为一项重要政治任务。为积极响应国家绿色低碳相关政策,改善碳排放工作中的相关问题,本文进行了如下研究:基于机器学习中无监督学习方法——K-Means聚类算法为企业构建碳排放审计预警模型,通过识别偏离的异常点,判定企业是否存在碳排放风险;并以我国重点钢铁企业为例进行了仿真;最后针对企业数据管理以及审计人员专业能力提出了两方面的保障措施,确保预警模型能够行之有效。
As China has put forward the goals of reaching peak carbon emissions and achieving carbon neutrality,the carbon emission audit has attracted more and more attention. China has always attached great importance to the data frauds of carbon emissions, and made it an important political task to crack down on data frauds and strengthen relevant supervision.In order to actively implement national policies of green and low-carbon development and solve relevant problems in the carbon emissions,this article researches how to use the K-Means clustering algorithm, an unsupervised learning method in machine learning, to construct an early risk warning model for carbon emission audit for the enterprises, which can identify the outliers with data skew sand judge whether the enterprises face risks in carbon emissions;it then conducts simulation test staking the key iron and steel enterprises as examples;finally, it puts forward the safeguard measures targeting both the data management inenterprises and the competence of auditors, so as to ensure the effectiveness of the early risk warning model.
出处
《中国注册会计师》
北大核心
2022年第12期14-20,I0002,共8页
The Chinese Certified Public Accountant
