Contributions to mathematical ranking models in learning analytics

Abstract

The emergence and rapid growth of the digital age has had an impact on nearly every aspect of our modern lives. Regarding tertiary education, teaching methods are changing, learning approaches are evolving and programs are becoming more accessible through the development of online learning. Although the obviously expected consequence should be an increased number of graduates and a more highly qualified generation of professionals, the student attrition rate at tertiary institutions has not improved. One of the reasons identified for this phenomenon is the observed academic complacency of enrolled students. Furthermore, many environmental, cultural, personal and psychological factors contribute to poor participation in academic activities. Studies confirm that inefficient feedback on their academic performance leaves students to belatedly realise that they will not be admitted to the final examinations. Learning at tertiary level needs to be improved by acknowledging its mutating character and implementing non-intrusive technology that embraces rather than rejects the unavoidable changes of the digital age. In this study, an academic performance feedback framework that utilises mathematical modelling techniques to facilitate an enhanced feedback approach for improved learning is proposed and evaluated. Comprehensive feedback contains information that reports on current academic performance (feed backwards) and is prescriptive in that it generates individualised improvement goals (feed upwards) and interim targets for reaching the ultimate improvement goals (feed forwards). It furthermore adheres to specific criteria to affect positive change. The literature study provides background on the current state of feedback in the educational environment, the definition and usefulness of learning analytics, mathematical modelling and how it can be applied to supplement learning analytics techniques, and frameworks. The attributes of feedback to be effective towards improved learning, were identified for incorporation into the framework. Learning analytics furthermore, presents valuable tools for managing and monitoring student participation in academic activities. However, it lacks some functionality in terms of calculating the academic performance of large groups of students and creating comprehensive feedback. Several different mathematical ranking modelling approaches were employed to first emulate, and then improve on existing academic performance calculation techniques. These approaches included various non-linear programming models, a data envelopment analysis model, the analytic hierarchy process, linear programming models and a decision tree approach. In order to generate comprehensive academic performance status reports that conform to the criteria for effective feedback, algorithms that implement a linear programming model, a non-linear programming model and a decision tree approach, were developed. A computerised demonstrator that incorporated the developed algorithms was created so as to establish the architecture required for use. The program was ultimately implemented and deployed in a specialised learning management system (SLMS), supplementing existing academic performance feedback. The SLMS implemented the improved mathematical models to dynamically create student academic performance feedback and was deployed parallel to an existing feedback approach, in a tertiary education environment. Evaluation of the SLMS was performed in a field test that provided some valuable insights to be considered for development of an academic performance feedback framework in tertiary education. Accordingly, a model-based framework was developed for academic performance feedback. The framework consisted of concepts and models, with a decision support system included as a learning analytics supplemental tool. The framework was employed and evaluated and yielded a very high degree of student satisfaction, improved program management by lecturers and institutions, accurate decision support in learning analytics and enhanced communication between institutions and students. Mass deployment of the framework will contribute to tertiary education by facilitating improved techniques to create effective and comprehensive academic performance feedback. Students will consequently be better equipped to make informed decisions regarding their conduct towards academic progress.