Intelligent performance cockpit for students (ASSIST)
In the ASSIST project, an AI-supported feedback portal is designed and tested as an intelligent assistance system for students. In this portal, students are shown their student performance in absolute terms and in relation to their cohort as well as their probability of success. In addition, students are shown the effects of performance in the following semester in CP and grade as well as passed specific exams on the probability of success in various scenarios. If students confirm their inclination to change, the effects of a change of program on the probability of success are also displayed in various predefined and self-selected scenarios. The portal is intended to provide students with an early decision-making basis for the further course of their studies by revealing their future course and the efforts required for successful completion of their studies.
The determination of success probabilities is based on the previous project FragSte, which was funded by the BMBF. FragSte is an early warning system that uses machine learning methods. It was designed, implemented and tested using data from two universities. In order to determine the effects of individual examinations and changes of study program, the already available mass data of students of the enrollment cohorts 2007-2020 of both universities have to be analyzed. For this purpose, different methods, especially event history analysis and collaborative filtering recommender systems, are to be used. The effects determined in this way are then to be integrated into the feedback portal.
Finally, the portal will be offered in a one-year beta phase to a stratified randomly selected group of students for voluntary use (experimental group). In the test phase, the behavior and performance of the students in the experimental group will be compared with those of the control group.
Datenschutzhinweise & Impressum
2021 until 2024
FragSte, FragSte 2, ASSIST, feedback, study dropouts, study progress, monitoring
Kerstin Schneider (project lead)
Simon Görtz (project lead)
Johannes Berens (contact person)
J. Leandro Henao