Hierarchical Clustering in Profiling University Students for Online Teaching and Learning

Data analytics provide an important contribution to improving educational processes, as well as managerial and organisational processes in higher education. They are also significant in transforming behavioural, performance, and interaction data on digital learning platforms into pivotal information about the learning process. This chapter presents analysis results from instructional and behavioural formative assessment data for an online common course stored on a Turkish state university’s learning management system using cluster analysis. This chapter focuses on evaluating activity and assessment data to discover student groups as an important input of assessment analytics. Developing supervised models require a priori labels, such as individual student success or fail scores, and a multidimensional dataset identifying relevant performance information gathered during the learning process. In a more realistic setting, the labels are not fully known until the semester end and supervised learning models may then create learning bias when trained only using assessment data. Therefore, cluster analysis presents a promising approach through observing performance similarities between students not yet evaluated for a course or subjected to end-of-term assessment. Although feedback mechanisms were not studied in this chapter, this approach can further help achieve necessary steps to increase student success before a course ends. This study grouped students based on learning performance data using cluster analysis, with models developed and evaluated according to various performance indicators. The main concepts related to the function of the methods in assessment analytics are discussed, and descriptive results obtained through the model are shared and evaluated for a specific case.