Project overview
The project entitled “Data-Driven Retention and Outcome Predictor (DROP)” (project ID: 2024-02-HU01-KA220-HED-000295434) aims to implement a coordinated, comprehensive development initiative across partner institutions. As a result of this collaboration, the participating universities will develop the capacity to predict student dropout and enhance their activities that support student success and prevent attrition.
Student dropout is a challenge faced not only by the partner institutions, but it is also considered a Europe-wide issue. Although numerous important scientific findings have emerged over the past decade and a half regarding the causes of dropout, the proportion of students leaving higher education has remained stable at 35–40%, indicating no significant reduction. The partner institutions are committed to tackling dropout, and their experience shows that traditional tools alone are insufficient. Effective reduction requires new, innovative, and data-driven solutions.
The primary objective of the DROP project is to develop a dropout-prevention methodology and intervention model, implemented through two key stages.
First, using the administrative data available at the institution and applying data mining methodologies, the project will identify students at risk of dropping out. With the support of the predictive models developed within the project, institutions will be able to establish a digital early warning system capable of signalling, in real time and at the individual student level, who is most at risk.
Second, continuing to build on administrative data, the project will provide personalised interventions that directly target the factors potentially causing dropout (e.g., difficulties in academic or social integration, lack of career commitment or orientation, financial challenges, etc.). This approach enables a shift away from the traditional dropout-prevention practices that rely heavily on students’ proactive help-seeking, replacing them with a proactive institutional support model.




