Student attrition was typically viewed from the perspective of psychology. In the past, those who did not stay in college were considered less talented, less motivated, and less willing to make the effort to delay earning college degrees [4]. Thus, the retention of students or the college's inability to retain them was linked to the individual skills, abilities, and motivation of the student. That is, if any students dropped out of college in their first year of enrollment, we said that the students failed, not the institutions. It is what we now know as victim-blaming [4].
To improve the quality of human resources, the development of the education sector is very important. Knowledge is power. The student enters the institution with many dreams and expectations. Therefore, the greatest responsibility of the institutions should be to fulfill the dreams of each student. For that, all the necessary requirements must be planned and organized with a defined learning pathway. Hence, the institution should be equipped with a tool to analyze potential student dropouts early in the semester and provide the appropriate support that is needed well in advance.
Significant studies have been developed applying Machine Learning in Higher Education based on concern for the student, with a specific focus on student academic performance, at-risk, and attrition [5]. To predict student performance and identify at-risk students, most papers use traditional Machine Learning algorithms, for instance, logistic regression, 𝑘-nearest neighbors, and decision tree-based ensemble models [6, 7]. Similarly, to predict student dropout and retention, the Naive-Bayes classification algorithm [8] and Support Vector Machines [9] have been applied, respectively.
On the other hand, the student’s demographic and socioeconomic aspects [10-12], academic history as well as admission test scores [6, 13, 14] have been shown to be key variables to predict the student dropout at Higher Level [8, 9]. Other factors were also found to have incremental predictive power on academic performance and retention, such as first-semester university performance indicators [15] and psychological factors [16]. Furthermore, dimensionality reduction techniques have been used to identify the main factors that affect early dropout [17].
The importance of having an accurate model to predict student dropout is that its results could be used to improve and develop retention strategies in Universities [9]. This would benefit the students, by having timely and personalized strategies from their Institution that support their permanence in their career, as well as the Institution, by improving their statistics of student degree completion and their student investment costs.
Therefore, the purpose of this Call for Proposals is to implement new solutions that allow predicting the dropout of a student in a Higher Institution by using Machine Learning models based on a curated educational dataset. The accepted proposals will be invited to submit an article for inclusion in a Special Issue of a high-impact Journal that we are preparing for this call. The article will be subject to meeting Journal’s scope and quality requirements.
Audience: This call is open to all researchers, faculty, analysts, graduate students with an interest in educational data and knowledge in Machine Learning algorithms. Joint proposals are welcome, with at least one researcher.
Keywords: Educational Innovation; Student Dropout; Student Attrition; Machine Learning; Data Hub; Higher Education.