Search

Modelling students' performance is one of the most challenging and exciting research areas of data analytics in higher education. This performance is influenced by multiple factors and is not accurately measured by the traditional five-point scale thus making this field attractive. The widespread availability of educational datasets further catalyses its popularity. In the present work, the authors investigate the performance of students starting their studies in the autumns of 2011, 2012, and 2013 at Corvinus University of Budapest, in the undergraduate business informatics training program, using variables available from the Neptun Education System. To accurately measure student performance, a new metric was developed that describes student performance on a broader set of values than the traditional five-point grade. This new metric is used to investigate the correlations between students' performances in subjects, applying structural equation models (SEM). The results show that through SEM, substantial relationships can be revealed between the investigated courses. A proper, interdependent system of prerequisites in the curriculum may be developed based on these results. The methodology presented in this study can also be used to anticipate possible difficulties students may face in terms of certain subjects and to offer tailored help. [ABSTRACT FROM AUTHOR]