Beijing Academy of Educational Sciences, Beijing 100036, China.
College of Teacher Education, Capital Normal University, Beijing 100048, China.
Int J Environ Res Public Health. 2022 Jul 31;19(15):9407. doi: 10.3390/ijerph19159407.
The emergence of the COVID-19 pandemic has hindered the achievement of the global Sustainable Development Goals (SDGs). Pro-environmental behaviour contributes to the achievement of the SDGs, and UNESCO considers college students as major contributors. There is a scarcity of research on college student pro-environmental behaviour and even less on the use of decision trees to predict pro-environmental behaviour. Therefore, this study aims to investigate the validity of applying a modified C5.0 decision-tree model to predict college student pro-environmental behaviour and to determine which variables can be used as predictors of such behaviour. To address these questions, 334 university students in Guangdong Province, China, completed a questionnaire that consisted of seven parts: the Perceived Behavioural Control Scale, the Social Identity Scale, the Innovative Behaviour Scale, the Sense of Place Scale, the Subjective Norms Scale, the Environmental Activism Scale, and the willingness to behave in an environmentally responsible manner scale. A modified C5.0 decision-tree model was also used to make predictions. The results showed that the main predictor variables for pro-environmental behaviour were willingness to behave in an environmentally responsible manner, innovative behaviour, and perceived behavioural control. The importance of willingness to behave in an environmentally responsible manner was 0.1562, the importance of innovative behaviour was 0.1404, and the perceived behavioural control was 0.1322. Secondly, there are 63.88% of those with high pro-environmental behaviour. Therefore, we conclude that the decision tree model is valid in predicting the pro-environmental behaviour of college student. The predictor variables for pro-environmental behaviour were, in order of importance: Willingness to behave in an environmentally responsible manner, Environmental Activism, Subjective Norms, Sense of Place, Innovative Behaviour, Social Identity, and Perceived Behavioural Control. This study establishes a link between machine learning and pro-environmental behaviour and broadens understanding of pro-environmental behaviour. It provides a research support with improving people's sustainable development philosophy and behaviour.
新冠疫情的出现阻碍了全球可持续发展目标（SDGs）的实现。 亲环境行为有助于实现可持续发展目标，教科文组织认为大学生是主要贡献者。目前，关于大学生亲环境行为的研究很少，甚至更少使用决策树来预测亲环境行为。因此，本研究旨在调查应用修改后的 C5.0 决策树模型来预测大学生亲环境行为的有效性，并确定哪些变量可用作此类行为的预测指标。为了解决这些问题，中国广东省的 334 名大学生完成了一份包含七个部分的问卷：感知行为控制量表、社会认同量表、创新行为量表、地方感量表、主观规范量表、环境行动主义量表和愿意以负责任的方式行事的意愿量表。还使用了修改后的 C5.0 决策树模型进行预测。结果表明，亲环境行为的主要预测变量是愿意以负责任的方式行事、创新行为和感知行为控制。愿意以负责任的方式行事的重要性为 0.1562，创新行为的重要性为 0.1404，感知行为控制的重要性为 0.1322。其次，有 63.88%的人具有较高的亲环境行为。因此，我们得出结论，决策树模型在预测大学生亲环境行为方面是有效的。亲环境行为的预测变量按重要性顺序排列为：愿意以负责任的方式行事、环境行动主义、主观规范、地方感、创新行为、社会认同和感知行为控制。本研究在机器学习和亲环境行为之间建立了联系，并拓宽了对亲环境行为的理解。它为改善人们的可持续发展理念和行为提供了研究支持。
