Determining Student Graduation Patterns Based on Semester Package Service Types at Universitas Terbuka Using the Apriori Algorithm

Figure Ilustration AI

FORMOSA NEWS - Palembang - Unlocking Student Success: How Data Mining Reveals Graduation Patterns in Distance Education. Researchers at Universitas Bina Darma and Universitas Terbuka have uncovered new insights into how student service packages influence graduation success in distance education. Published in 2026 within the Formosa Journal of Computer and Information Science, the study analyzed historical academic records to help higher education institutions improve student monitoring and administrative services.

Background and Research Methodology

Modern universities accumulate vast amounts of digital records through student registration, online learning platforms, and graduation tracking. While these databases are essential for daily administration, they also offer hidden opportunities to discover recurring behavioral trends through educational data mining. At Universitas Terbuka, an open and distance higher education institution characterized by flexible, geographically distributed learning, understanding these patterns is vital for optimizing student support. To investigate these dynamics, researchers Lora Ersitha, Yesi Novaria Kunang, Usman Ependi, and Ari Muzakir analyzed a dataset of 1,134 alumni from the Bachelor of Management program who graduated between 2020 and 2025 at Universitas Terbuka Palembang. The team transformed student records into categorical transaction items examining four key attributes.

• Semester Package Service: Ranging from structured options like SIPAS Semi, SIPAS Non-TTM, and SIPAS Penuh to flexible course registration options like Non-SIPAS.
  
• Study Completion Status: Categorized by timely completion versus going beyond the standard study period using learning management systems.
  
• Graduation Predicate: Classifications including Cum Laude, Very Satisfactory, and Satisfactory.
  
• Learning Support: Engagement attributes linked to learning management system utilization.
  

The researchers implemented the Apriori association rule algorithm using RapidMiner software to detect recurring item combinations. The analysis utilized a minimum support threshold of 10% and a minimum confidence threshold of 70%, evaluating rule strength using support, confidence, and lift metrics.

Key Findings and Patterns
The Apriori analysis successfully generated eight primary association rules that met the study's strict statistical thresholds.

• The Most Dominant Pattern: The broadest trend linking academic success showed that timely study completion frequently co-occurs with a "Very Satisfactory" graduation predicate, appearing in 42% of the analyzed transactions with 77% confidence.
  
• Single-Service Trends: Among individual service schemes, SIPAS Non-TTM paired with a Very Satisfactory predicate yielded the highest single-service support at 25%.
  
• The Most Representative Service Rule: When combining service models with completion timelines, the combination of timely completion and SIPAS Semi leading to a Very Satisfactory predicate emerged as the most informative service-related rule, achieving 15% support, 81% confidence, and a positive lift of 1.13.
  
• Learning Management System Insights: The data revealed a strong positive association between a Satisfactory graduation predicate and learning management system occurrence, yielding the highest lift value of 1.66 among all primary rules.
  

Real-World Impact and Implications
The findings provide actionable intelligence for academic administrators, student advisors, and institutional policymakers in open-learning environments. Rather than viewing student data as mere archives, universities can utilize association rule mining as a descriptive diagnostic tool. By recognizing which service combinations frequently align with timely graduation and strong academic standing, institutions can refine tutorial support structures and target early interventions for student groups exhibiting delayed completion risks.

Author Profile
Lora Ersitha: Master of Computer Science, Universitas Bina Darma, and academic staff at Universitas Terbuka Palembang. Her field of expertise includes educational data mining, information systems, and data analytics.
Yesi Novaria Kunang: Doctor of Philosophy, Universitas Bina Darma. Her field of expertise covers computer science, algorithm optimization, and information security.
Usman Ependi: Doctor of Computer Science, Universitas Bina Darma. His field of expertise encompasses information systems evaluation and educational technology.
Ari Muzakir: Master of Computer Science, Universitas Bina Darma. His field of expertise includes computational intelligence and data processing applications.

Source
Lora Ersitha, Yesi Novaria Kunang, Usman Ependi, Ari Muzakir. Determining Student Graduation Patterns Based on Semester Package Service Types at Universitas Terbuka Using the Apriori Algorithm. Formosa Journal of Computer and Information Science (FJCIS). Vol. 5, No. 2, Halaman 445-456.
DOI : https://doi.org/10.55927/fjcis.v5i2.17332
URL: https://journal.formosapublisher.org/index.php/fjcis: