Article Preview
Top1. Introduction
Despite the recent growth of online education under the paradigm of Technology-enhanced learning or TEL (Cheng et al., 2021; Waheed et al., 2020; Rajabi & Greller, 2019), traditional universities remain the primary source of education for the masses. Academic performance improvement and early intervention of at-risk students’ remains a challenging task in any educational setting (Fayoumi & Hajjar, 2020). Our research provides prediction-based models on the data taken from student information system by tapping the power of machine learning (Jiang, Gradus, and Rosellini 2020) to predict the academic performances of students with high accuracy and those at-risk of graduating late. The employed models assist instructors in forming appropriate pedagogical intervention strategies for optimal resource allocation of an institute (Maheshwari et al. 2020). Overall, the increased recognition of online learning platforms has yielded a progression in the data repositories about students' interactions and online activities, resulting in several educational research communities (Aldowah, Al-Samarraie, and Fauzy 2019; Avella et al. 2016).
In contrast to the online learning data repositories and their analysis in predicting students' academic performances, little work has been conducted on the students’ interactions and influential attributes impacting their academic performances in traditional classroom settings (George & Lal, 2021). This lack of student engagement data, coupled with the reluctance of universities to share their data due to privacy concerns, becomes a hindrance in determining student academic performance and identifying at-risk students, especially in traditional classroom settings. In the existing studies, analyzing and predicting the performances of students has received considerable attention in the educational data mining community and hence in the newly emerging related fields, such as learning analytics, this particular objective has evolved in terms of early identifying the student at-risk of low performances, during an ongoing course (Chanlekha and Niramitranon 2018; Hassan et al. 2019).
Furthermore, another dimension prevalent in par with predicting academic performances is analyzing the student's time to graduate a degree. Learning analytics also emphasizes the optimal resource allocation of an institute for strategizing the administrative tasks, regulating performances and maintaining learning resources in higher education (Waheed et al. 2018). The capacity of an institute is also a significant predictor in analyzing the resource maintenance mechanism for more optimal allocation.
To assist the educational stakeholders in forming instructional pedagogical interventions, improving the academic performances of students and identifying the students at-risk of low performances and at-risk of graduating late from the institute, this study leverages machine learning techniques to analyze these perspectives of an institute. The research objectives addressed in this study intend to leverage the student data from traditional classroom settings for a more thorough analysis of student behavior, and are stated as follows: