A Comparative Analysis between K-Means and Agglomerative Clustering Techniques in Maritime Skill Certification

Abstract

The maritime industry must constantly adjust seafarer training to meet evolving operational demands and ensure compliance with new regulations. This study addresses the challenge of assessing the relevance of Certificate of Proficiency (COP) services by categorizing them to determine which qualifications are essential for marine professionals. The goal is to identify obsolete or misaligned training programs that need updates or enhancements to better serve industry needs. To this end, the study employed two clustering algorithms, K-Means and Agglomerative Clustering, on data from 2021 to 2023. K-Means was chosen for its efficiency in processing large datasets and creating clear, non-overlapping groups. Agglomerative Clustering was selected for its ability to offer a detailed, hierarchical view of data, which helps in understanding the complex structure of certification demands more comprehensively. The analysis identified three main clusters; notably, Cluster 2 indicated a high demand for critical certifications, while Cluster 1, containing the majority of certifications, received little interest, suggesting they may be less relevant. This insight encourages training providers to consider refining their offerings. Although comprehensive, the study's three-year timeframe suggests extending this period in future research for a more detailed trend analysis and forecasting in maritime training adaptations.

Keywords

• maritime skill certification clustering k-means agglomerative data mining