<p>National mathematics assessments provide valuable insights into student learning but often rely on broad proficiency classifications that fail to capture detailed cognitive strengths and weaknesses. Therefore, this study analyzed students’ mastery of cognitive attributes using data from South Korea’s National Assessment of Educational Achievement. Association rule mining was employed to explore the hierarchical relationships among cognitive attributes for refining learning progression and supporting personalized instruction. The findings suggested that representing mathematical ideas is the first cognitive attribute that must be mastered, followed by conceptual understanding, reasoning, transforming representations, and problem-solving in order. The results also revealed prerequisite relationships among skills that are not explicitly recognized in existing curricula but emerge through student performance patterns. These insights will contribute to refining instructional strategies and designing targeted interventions. By integrating cognitive diagnostic insights into large-scale assessments, educators can provide precise and personalized feedback and develop adaptive learning strategies to support student growth. Despite being limited by the test specifications of the National Assessment of Educational Achievement, this study enhances the understanding of cognitive structures in mathematics achievement and highlights the need for further empirical validation of mastery hierarchies. The findings emphasize the importance of leveraging data-driven approaches to improve mathematics education and ensure that assessment outcomes translate into actionable instructional improvements.</p>