Scholarlink

<p>This paper compares the educational performance of a set of countries based on the PISA 2022 results in Mathematics, Reading, and Science. The study adopts a multianalytical quantitative framework that integrates classical statistical techniques with unsupervised and supervised machine learning. Methods correlation analysis, Principal Component Analysis, and regression. The results of the regression imply that mathematics is the strongest predictor of general PISA achievement. Country groupings are determined using K-Means clustering, while the robustness of these clusters is checked by performing several runs of supervised classification models (Random Forest, Extra Trees, Support Vector Classifier, and Multilayer Perceptron) embedded in a stratified k-fold cross-validation framework. The findings indicate a consistent high achievement of East Asian education systems, a relatively balanced situation in many European countries, and a persistent low achievement in some regions. Cluster-based results from a policy point of view allow one to conclude that strengthening mathematical literacy, improving teacher quality, and pursuing curriculum coherence are the most powerful levers in reducing international differences in learning outcomes. Given that this study translates empirical cluster structures into targeted, data-driven reform priorities, actionable guidance for context-sensitive, evidence-based education policy design follows.</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

Comparative analysis of global education performances based on PISA 2022 results: an assessment with statistical and machine learning methods

  • 期刊论文

  • 发表日期：

    2026年6月16日

查看全文
  • Sadullah Çelik,
  • Bilge Doğanlı,
  • Emel Şaşmaz

摘要

This paper compares the educational performance of a set of countries based on the PISA 2022 results in Mathematics, Reading, and Science. The study adopts a multianalytical quantitative framework that integrates classical statistical techniques with unsupervised and supervised machine learning. Methods correlation analysis, Principal Component Analysis, and regression. The results of the regression imply that mathematics is the strongest predictor of general PISA achievement. Country groupings are determined using K-Means clustering, while the robustness of these clusters is checked by performing several runs of supervised classification models (Random Forest, Extra Trees, Support Vector Classifier, and Multilayer Perceptron) embedded in a stratified k-fold cross-validation framework. The findings indicate a consistent high achievement of East Asian education systems, a relatively balanced situation in many European countries, and a persistent low achievement in some regions. Cluster-based results from a policy point of view allow one to conclude that strengthening mathematical literacy, improving teacher quality, and pursuing curriculum coherence are the most powerful levers in reducing international differences in learning outcomes. Given that this study translates empirical cluster structures into targeted, data-driven reform priorities, actionable guidance for context-sensitive, evidence-based education policy design follows.