Educational data mining: factors influencing medical student success and the exploration of visualization techniques

Objectives Medical student education is critical in equipping future doctors to impact patient healthcare and the national public health system significantly. This study aimed to identify factors influencing student academic success (honors level or high-grade group) among medical students using data mining techniques applied to multidimensional educational data. Materials and methods A retrospective cohort study was conducted using a standardized questionnaire administered to 145 medical students. A total of 13 factors spanning four domains—academic activity, demographics, environment, and psychology or learning style—were examined. The prevalence ratio (PR) and adjusted prevalence ratio (APR) were calculated using multivariate logistic regression. Unsupervised learning techniques, including cluster analysis and association rules, were used to identify hidden patterns. Visualization techniques, such as heatmaps and centroid plots derived from cluster analysis, were employed to depict data relationships and facilitate the interpretation of key trends. Internal validation was also evaluated. Results Among the 13 factors analyzed, logistic regression identified a pre-med GPAX ≥3.75 and an interest in internal medicine as statistically significant predictors of high academic performance, with adjusted prevalence ratios (APRs) of 1.73 (95% CI, 1.02–2.91, p = 0.040) and 1.52 (95% CI, 1.14–2.03, p = 0.005), respectively. Cluster analysis revealed characteristic traits of high-grade students, including metropolitan residence, very high pre-med GPAX, and a preference for kinesthetic and reading learning styles. Association rules analysis further emphasized the importance of environmental factors, particularly transportation time to school and access to learning resources, in supporting academic success. Conclusion Educational data mining (EDM) provided valuable insights into factors contributing to medical student success. Logistic regression highlighted pre-med GPAX and an interest in internal medicine as key predictors. Cluster analysis uncovered patterns linking learning styles and academic performance, while association rules emphasized the role of environmental factors, such as school proximity and resource availability. Together, these methods provide a comprehensive and visually engaging framework to inform educational planning, potentially generating novel insights for addressing medical challenges and enhancing clinical practice.