学习环境对中学生全球素养的影响机制与循证决策研究：基于机器学习的关联规则挖掘

全球素养是近年来核心素养运动的重要维度。学习环境对学习结果的作用路径复杂，由于教育统计局限，未知与隐含学习环境因素之关联互动的作用尚需揭示，学习环境对全球素养的影响机制亟待解决。前期研究发现，学习环境影响效应研究经历了提取影响因素、层级分析到交互结构三个阶段，基于单一或多维的中介关系进行假设验证。我们认为学习环境与其他因素交互作用于全球素养，机器学习的数据挖掘将成为方法瓶颈的突破口。本课题首先基于学习环境及其影响效应的元分析，从理论上构建全球素养支持模型；其次，基于PISA2018大数据进行学习环境特征选择与全球素养关联关系的全面获取，优化机器学习模型并确定核心影响因素；再次，基于树模型与Apriori关联分析的规则挖掘，分析核心影响因素的逻辑关系，并发现学习环境交互网络的影响机制。最后，通过三角互证构建预测模型，结合政策案例，提出循证决策建议。这将有助于全球素养教育实施和学习环境治理。

Global competence has become an important dimension of core competence movement in recent years. Due to the complicated influencing mechanism of learning environment on learning outcomes as well as the limitation of educational statistics, the complex interaction between the unknown and the implicit learning environment factors has not been revealed. Therefore, the influencing mechanism of learning environment on global competence needs to be identified urgently. Our previous research found that the study of the effects of learning environment has undergone three stages: identifying influencing factors, analyzing hierarchically and interactive structure, based on a single or multi-dimensional intermediary relationship in a research from hypothesis to verification. We believe that learning environment and other factors interactively influence global competence, and data based on machine learning is the breakthrough in methods. Firstly, based on the meta-analysis of learning environment and its influence, this research constructs a support model of global competence in theory. Secondly, based on the data of PISA 2018, the research attempts to comprehensively acquire the features of learning environment and its relationship with global competence, optimize the model of machine learning and determine the core influencing factors. Thirdly, based on the Tree model and Apriori association analysis, the logical relationship of the core influencing factors is analyzed, and the influencing mechanism of learning environment is also identified. Finally, this research proposes some evidence-based policymaking suggestions by constructing a prediction model based on the triangle method and analyzing policy cases. This will contribute to the implementation of global competence education and the improvement of learning environment.