Abstract:

Higher education institutions (HEIs) collect a great amount of data on students, including their academic, demographics and activity information. At the same time, the use of student data for predictions in many HEIs is very limited or non-existent. Educational data mining (EDM) can be used for early identification of the actual situation with each student and enhance student learning and achievements, enable individualized teaching as well as strengthening well-informed communication between students and their support personnel. These opportunities are not used due to unavailability of an appropriate analytical ecosystem.

According to several authors, student dropout is considered the most complex and significant issue in the education system [1]. It causes economic, social, academic, political, and financial issues not only for the learner but for society in general [2]. Early student dropout prediction is one of the problems addressed by EDM. Meanwhile, research to date in student success analysis, especially dropout analysis is rather simplistic, considering the limited data set, limited factors and limited toolset [3]. Commercial tools working in the field of student success analytics, like SEAtS and various plugins in Moodle, have limited capabilities, are tied to specific vendors and often lack the required flexibility due to varied state policies and education institution regulations. The state of the art in using data analytics for other issues in HEIs is even less developed.

The long-term aim of the research is to create a data analytics ecosystem that serves the stakeholders in higher education by providing the necessary knowledge. In our vision, the solution would take the existing data about students as input, use the knowledge from previous cohorts and provide live information about the progress of each student.

The paper is the first step towards such a solution. The goal of the paper is to use the existing situation at HEIs as a basis to define already existing elements and missing ones to form the necessary ecosystem that would enable the use of education data analytics to deliver the necessary knowledge for lecturers, administrators and students themselves. The paper presents:
(1) analysis of relevant data sources,
(2) an analytical process using machine learning and
(3) provisioning stakeholders with relevant information through customized dashboards.

The development of machine learning models and the overall analytical pipeline for analysing student success forms a basis for delivering analytical support to stakeholders at a HEI and facilitating individualized learning experiences. The purpose of tailoring the educational process to adapt to various needs of struggling or talented students and timely intervene with students at risk.

References:
[1] Kim D., Kim S. (2018). Sustainable Education: Analyzing the Determinants of University Student Dropout by Nonlinear Panel Data Models. Sustainability. Vol. 10(4):954.
[2] Niyogisubizo, J., Liao, L., Nziyumva, E., Murwanashyaka, E., & Nshimyumukiza, P. C. (2022). Predicting student’s dropout in university classes using two-layer ensemble machine learning approach: A novel stacked generalization. Computers and Education: Artificial Intelligence, 3, 100066.
[3] Mariano, A. M., Ferreira, A. B. de M. L., Santos, M. R., Castilho, M. L., & Bastos, A. C. (2022). Decision trees for predicting dropout in Engineering Course students in Brazil. Procedia Computer Science, 214, 1113–1120.

Keywords:

Learning Analytics, Data Mining, Machine Learning, Student Support, Higher Education Institutions.