<![CDATA[The utilization of academic data in educational institutions is often underutilized, leading to the "Data Rich, Information Poor" phenomenon. At SMK Negeri 2 Ulu Moroo, student ability evaluation still relies on a single linear average value, which contains methodological flaws as it disguises the disparity between theoretical competence (cognitive) and practical vocational skills (psychomotor). To address this issue, this research applies Educational Data Mining (EDM) techniques using the K-Means Clustering algorithm based on the CRISP-DM framework. The study involved $N = 176$ students from grades X and XI, with feature variables including Cognitive ($X_1$), Psychomotor ($X_2$), and Affective ($X_3$) scores, which were homogenized using Min-Max Normalization $[0, 1]$ to eliminate scale bias. Cluster validation using a combination of the Elbow Method and Silhouette Coefficient determined the optimal number of clusters at $k = 3$, with a WCSS variance reduction of $59.95\%$ and Silhouette score of $0.68$ (Strong Structure). Centroid denormalization partitioned student academic profiles into three categories, namely Cluster 1 or High Achievers consisting of 62 students showing linear dominance with cognitive score of $86.45$ and a psychomotor score of $89.20$, Cluster 2 or Middle Achievers with 79 students having a stable cognitive score of $74.20$ but a fluctuating psychomotor score of $78.10$, and Cluster 3 or Underachievers comprising 35 students performing below the passing grade with a cognitive score of $68.70$ and a psychomotor score of $66.40$. These findings serve as a Decision Support System the school to implement differentiated learning, targeted remedial programs, and evidence-based industrial internship placements.]]>
Copyrights © 2026
