Using Clustering to Interpret PISA Mathematics Performance: A Data-Driven Insight for Mathematical Learning and Equity

This Data Note demonstrates how a transparent data-driven approach can be used to interpret patterns of mathematical learning outcomes across countries. Using K-Means clustering on country-level mathematics mean scores from the Programme for International Student Assessment (PISA) 2015, 2018, and 2022, three stable groups emerged: High Performers (approximately 514), Moderate Performers (approximately 420), and Low Performers (approximately 370), with acceptable validity (Davies-Bouldin Index = 0.67). These values represent country-level cluster mean scores, not the number of countries in each cluster or official student-level PISA proficiency categories. The clustering approach enables educational researchers, graduate students, and instructors of educational statistics or comparative education to visualize similarities and disparities among countries, thereby revealing non-linear performance patterns and cross-country similarities that are not captured by ranking-based reporting. The results highlight persistent learning gaps associated with levels of national development while also showing exceptions that invite closer comparative interpretation. The method is replicable, low-cost, and adaptable for research, graduate-level teaching, or methodological demonstration.