Kerstin Wagner

In this paper, we present an extended evaluation of a course recommender system designed to support students who struggle in the first semesters of their studies and are at risk of dropping out. The system, which was developed in earlier work using a student-centered design, is based on the explainable k-nearest neighbor algorithm and recommends a set of courses that have been passed by the majority of successful neighbors, that is, students who graduated from the study program. In terms of the number of recommended courses, we found a discrepancy between the number of courses that struggling students are recommended to take and the actual number of courses they take. This indicates that there may be an alternative path that these students could consider. However, the recommended courses align well with the courses taken by students who successfully graduated. This suggests that even students who are performing well could still benefit from the course recommender system designed for students at-risk. In the present work, we investigate a second type of success - a specific minimum number of courses passed - and compare the results with our first approach from previous work. With the second type, the information about success might be already available after one semester instead of after graduation which allows faster growth of the database and faster response to curricular changes. The evaluation of three different study programs in terms of dropout risk reduction and recommendation quality suggests that course recommendations based on students passing at least three courses in the following semester can be an alternative to guide students on a successful path. The aggregated result data and results explorations are available at: https://kwbln.github.io/jedm23.

In this work, we present a survey of a course recommender conducted among students and its results. The course recommender system, published in our previours work (Wagner et al., 2023), is based on the nearest neighbors algorithm and aims to support students in their course enrollment; it targets above all students who did not pass all mandatory courses as indicated in the study handbook in their first or second semester at university. The primary objective of the survey was to evaluate the perceived quality of explanations and recommendations based on two presentation variants (a ranked list of courses and a set of courses), as well as the general trust in such systems. The survey included quantitative measures and demographic information from the students, so that different subgroups could be evaluated. The results indicate that students tend to trust recommender systems and that they tend to understand the explanations. No clear winner emerges between the presentation of the courses as a set and as a ranked list. The survey data explorations are available at: https://kwbln.github.io/csedu24.

In this paper, we present an extended evaluation of a course recommendation system to primarily support students who struggle in the first semesters of their studies and are at risk of dropping out. The course recommendation system was developed in earlier work using a student-centered design and is based on the explainable k-Nearest Neighbor algorithm. The system recommends the set of courses that have been passed by the majority of their nearest neighbors who have completed their studies. The present evaluation is based on the data of students from three different study programs. The first result is that the recommendations do lower the dropout risk. A second result is that the set of recommended courses matches quite well the courses taken by students who completed the study program but differs from the courses taken by struggling students. Thus, though the course recommended system targets primarily struggling students, students doing well could use it. A third result is that the number of recommended courses for struggling students is less than the number of courses they actually enrolled in. We state that the recommendations given indicate a different and hopefully feasible path through the study program for students at risk of dropping out.

To predict whether students will drop out of their degree program in a middle-sized German university, we investigate five algorithms - three explainable and two not - along with two different feature sets. It turns out that the models obtained with Logistic Regression (LR), an explainable algorithm, have the best performance. This is an important finding to be able to generate explanations for stakeholders in future work. The models trained with a local feature set and those trained with a global feature set show similar performance results. Further, we study whether the models built with LR are fair with respect to both male and female students as well as the study programs considered in this study. Unfortunately, this is not always the case. This might be due to differences in the dropout rates between subpopulations. This limit should be taken into account in practice.

Selecting courses that optimally fit a student’s situation can help reduce the risk of dropping out. Data exploration and performance prediction approaches can be applied to help students make these decisions. To ensure that an enrollment support system meets the needs of students, they should be involved as early as possible in the development process. This paper presents an initial assessment of some functionalities of a novel course enrollment support system based on student performance data. The results include a collection of indicators and sources of information, as well as an overview of needs and concerns. The insights gathered will help to develop a system that has the trust of students.