Enhancing teaching and learning through data-driven optimization of servicing code demand and lecturer allocation using WEKA analysis

The increasing demand for servicing codes across faculties has created a growing need for data-driven decision supports in optimizing lecturer allocation and cost efficiency. This study applies machine learning techniques using the WEKA analytical tool to explore, cluster and classify servicing code applications using a dataset gathered from multiple faculties, and campuses within the Faculty of Business and Management in a selected public university in Malaysia. The dataset of 297 instances comprised attributes such as Course Code, Course Name, Course Type, Faculty, Program, Campus, Total Number of Students Enrolled and Approval Status. The main objective of this study is to identify the demand patterns and optimizing the lecturer’s contribution by maintaining a class sizes of maximum number of students in each class is 30 and a teaching load of up to 20 credit hours per lecturer. An Expectation-Maximization (EM) clustering model revealed five distinct clusters representing varied course demand concentrations and faculty distributions, with Cluster 1 (30%) showing the highest cumulative demand across university courses. Complementary K-Means clustering grouped the data into two major clusters, indicating that a clear differentiation between economic-based and entrepreneurship-based courses in terms of student enrolment volume and approval distribution. Attribute selection through Information Gain Attrite Evaluation model highlighted Program Code, Course Code and Type of Course as the strongest predictors of course approval and demand levels. Furthermore, classification using the Random Forest algorithm depicted that a 95.3% accuracy (k=0.768), confirming robust predictive capability in identifying course approval status and demand trends. These results identifying course approval status and demand trends. These re suggest that machine learning driven approaches can effectively support academic administrations in making informed staffing decisions, balancing full time and part time lecturer assignments, and optimizing cost structures without compromising teaching quality. Theoretically, this study contributes to the emerging literature on data-driven academic resource management and the application of artificial intelligence in higher education operations. Practically, it offers a replicable analytical framework for institutions seeking to forecast servicing code demand and align lecturer allocation strategies with real time dynamics and cost optimization goals.