<![CDATA[Graduate work-readiness segmentation is essential for data-driven career services in universities. This study applies K-Means clustering to tracer-study data using four input indicators: GPA (IPK), TOEFL, soft-skill points (SSKM), and study duration, while employment status and waiting time are treated as external outcomes. Records from 669 graduates (2020–2023) were preprocessed via deduplication, range checks, and z-score standardization. The number of clusters was determined data-driven over K=2–10 using the Elbow Method (SSE) and Davies–Bouldin Index; the optimal K=9 was selected at the DBI minimum. PCA visualization indicated a distinguishable cluster structure. Clusters C0, C3, C5, and C7 exhibited faster transitions (median waiting time 2 months) with high employment proportions (up to ~90%), whereas C2 and C8 showed longer waiting times (≥4 months). Cluster C4 was characterized by the longest study duration and a comparatively lower employment proportion. These results demonstrate that unsupervised learning can reveal actionable readiness segments, supporting targeted interventions (e.g., CV/portfolio clinics, interview practice, structured internships) and providing a foundation for subsequent predictive modeling of graduate outcomes.]]>
Copyrights © 2026
