Using Unsupervised Learning to Identify Student Performance Profiles in English Language Education: A Clustering Analysis from the Nigeria Maritime University

The ability to communicate effectively in English is fundamental to academic success in Nigerian universities, yet persistent low achievement in the compulsory “Use of English” course remains a concern, particularly in specialized institutions like maritime universities where proficiency underpins both academic performance and professional competence. This study addresses the limitation of reactive academic support systems by employing unsupervised machine learning to analyze English language learning characteristics among undergraduates. Using K-means clustering on survey data from 248 students at the Nigeria maritime university, the research identified four distinct student performance clusters validated through chi-square analysis (p = 0.000949). The clusters revealed multidimensional profiles incorporating academic, psychological, and environmental factors: Cluster 1 (Diligent High-Achievers) demonstrated strong study habits and confidence; Cluster 2 (Steady Performers) represented average students with consistent patterns; Cluster 0 (Quiet Achievers) achieved comparable results through different behavioral pathways; and critically, Cluster 3 (At-Risk Group) exhibited low study hours, diminished confidence, elevated anxiety, and minimal AI tool adoption, with a 36.4% failure rate. The findings demonstrate that clustering techniques enable early identification of learning difficulties before traditional assessment methods detect poor performance. Recommendations include developing differentiated institutional support systems targeting specific cluster needs, from foundational programs for at-risk students to enrichment opportunities for high achievers, advancing data-driven approaches to language education in Nigerian higher education.