<![CDATA[Lecturer performance evaluation (EDOM) in higher education is often reduced to a single average score that fails to capture the heterogeneity of teaching competency profiles. This study deconstructs lecturer performance patterns through an Educational Data Mining (EDM) approach to overcome the "Data Rich, Information Poor" phenomenon. Using an eight-semester longitudinal database (2021–2024) covering a population of 493 lecturers and 5,577 evaluation records at UIN Sultanah Nahrasiyah, an adaptive scale-harmonization procedure was applied to equate evaluation instruments that differed across years. Empirical validation using the Elbow Method identified three distinct performance typologies: an Excellence cluster (32.8%; IKD 3.97, minimal load), a High Productivity cluster (36.1%; extreme load of 23.69 classes, moderate performance), and an Underperformance cluster (31.1%; fundamental competency constraints despite a low workload). Pearson and Spearman correlation tests confirmed no significant linear relationship between teaching load and IKD (p > 0.05), reinforcing the non-linear nature of the relationship. These findings confirm the existence of a Productivity Paradox in higher education and recommend a shift from uniform HR development policies toward a data-cluster-based differentiation strategy.]]>
Copyrights © 2026
