Classifying northern Italian students in their transition to master degree

The university students’ behaviour represents a relevant field of study from the management point of view. Given the availability of large administrative data on students’ careers, the chance to discover students’ profiles in terms of behavioural patterns could be interesting. However, the identification of students’ clusters that are informative, feasible and robust at the same time could be complex. The present work aims to define a feasible student clusterisation, adopting an empirical algorithm to treat mixed data and large sample sizes and borrow the syncytial clustering idea developed in the machine learning framework. The proposal is a generalisation of the original algorithm to mixed data cases. Finally, the importance of finding a prototype of students’ behaviours is discussed.