Abstract
This research examines the quality of academic services at Muhammadiyah University of West Sumatra (UMSB) as a strategy in responding to competitive dynamics in the higher education sector. The study aims to predict service quality and student satisfaction at Muhammadiyah University of West Sumatra by adopting a Machine Learning approach that combines predictive analysis methods of Linear Regression and Naive Bayes classification. The research utilized a primary dataset encompassing 2,000 student respondents through stratified random sampling technique. The novelty of this research lies in the integration of the SERVQUAL instrument with computational algorithms to generate deeper and more measurable identification of satisfaction factors. Research findings reveal variations in the level of dominance across the five measured dimensions. The Assurance, Reliability, and Empathy dimensions respectively occupy the top three ranks as the most dominant dimensions (average above 0.220). Meanwhile, the Tangibles and Responsiveness dimensions occupy the lowest ranks, with Responsiveness being the weakest aspect (0.146) and therefore requiring more attention for improvement. Model reliability was evaluated using accuracy, precision, recall, and F1-score metrics, as well as error measurement through RMSE and MAE. This research provides practical contribution in the form of strategic recommendations for university management to implement data-driven service improvements while serving as a foundational framework for the development of an integrated student satisfaction monitoring system within the academic information ecosystem.
