The Extension of Attachment Theory into School Contexts and Adolescent’s Social and Emotion Skills Development A Regional Comparative Study

This dataset contains the analytical materials supporting the study titled “Predictors of Adolescents’ Social and Emotional Skills: A Cross-Cultural Analysis Using Machine Learning and SHAP Interpretation.” The study used data from the OECD 2023 Survey on Social and Emotional Skills (SSES) to examine the key predictors of social and emotional skills among adolescents from different cultural contexts. The materials include data processing procedures, machine learning analysis codes, model development scripts, and SHAP-based interpretation procedures used in the study. The analysis involved comparisons of five machine learning algorithms, including XGBoost, Random Forest, LightGBM, CatBoost, and Linear Regression, with CatBoost selected as the optimal predictive model. SHAP analysis was conducted to evaluate feature importance and explore nonlinear relationships between predictors and students’ social and emotional skills. The original OECD SSES 2023 student-level dataset is not included in this repository due to OECD data access and redistribution restrictions. Researchers interested in reproducing the analyses should obtain the original dataset directly from the OECD following the relevant data access procedures. The materials provided here facilitate transparency, reproducibility, and further research on adolescent social and emotional development using interpretable machine learning approaches.本数据集包含支撑题为《青少年社会情感技能预测因素：基于机器学习与SHAP（SHapley Additive exPlanations）解释的跨文化分析》的研究的相关分析材料。本研究采用经济合作与发展组织（Organisation for Economic Co-operation and Development，OECD）2023年社会情感技能调查（Survey on Social and Emotional Skills，SSES）的数据，旨在考察不同文化背景下青少年社会情感技能的核心预测因素。 本次研究涉及的分析材料包括数据处理流程、机器学习分析代码、模型开发脚本以及基于SHAP的解释流程。本次分析对比了五种机器学习算法，包括XGBoost、随机森林（Random Forest）、LightGBM、CatBoost以及线性回归，最终选定CatBoost作为最优预测模型。本研究开展了SHAP分析，以评估特征重要性，并探究预测因素与学生社会情感技能之间的非线性关联。 由于经济合作与发展组织的数据获取与再分发限制，本仓库未包含原始OECD SSES 2023学生级数据集。有意复现本次分析的研究者需遵循相关数据获取流程，直接从经济合作与发展组织获取原始数据集。本数据集提供的材料有助于提升研究透明度、可复现性，并为使用可解释机器学习方法开展青少年社会情感发展相关的后续研究提供支撑。