<![CDATA[Lecturer performance evaluation is a crucial component in efforts to improve the quality of higher education. However, traditional evaluation methods still face various challenges, such as subjective assessments, a lack of consistent standards, and lengthy decision-making processes. These conditions highlight the need for a more measurable, accurate, and data-driven evaluation mechanism, particularly in the context of ongoing digital transformation. This study aims to design and develop a lecturer performance prediction system using a machine learning (ML) approach within a Decision Support System (DSS) framework. The research approach involves processing historical lecturer data covering aspects of Teaching (including student evaluation scores, instructional innovation, and attendance levels), Research (number of publications, H-index, and participation in academic conferences), Community Service, and other administrative activities. Predictive models were developed and compared using several machine learning algorithms, namely Random Forest, Support Vector Machine (SVM), Multilayer Perceptron (MLP), and XGBoost. Experimental results show that Random Forest achieved an accuracy of 88.0%, SVM 85.0%, and MLP 87.0%, while XGBoost demonstrated the best performance with an accuracy of 92.0%, precision of 91.0%, recall of 90.0%, and an F1-score of 91.0%. Based on these results, XGBoost was selected as the primary model for the DSS. In addition, the system is equipped with a rule-based module that generates follow-up recommendations based on the model’s prediction results. All system components are implemented in an interactive dashboard using the Streamlit framework, enabling users to input data, monitor prediction outcomes, and obtain decision recommendations in a fast and data-driven manner.]]>
