Learning analytics techniques and visualisation with textual data for determining causes of academic failure

ABSTRACT

The primary goal of higher education institutions is to support all students in the pursuit of academic success, which requires timely assistance for ‘at risk’ students. The adoption of learning management systems results in a large amount of data that can be collected, processed and utilised to improve the students’ learning experiences. This research examines the potential applications of analytics techniques for extracting insights from student-generated content in an academic setting. It showcases how different text analytics techniques, from descriptive content analysis, semantic network analysis, to topic modelling support the discovery of new insights from unstructured, user-generated data. We looked at 968 letters written by ‘at risk’ students in an Australian-based university in Southeast Asia to examine the difficulties the students faced, which led to their academic failure. The results show that time management, family, learning, assessment, and subjects were the leading causes of poor performance, but in a more nuanced way than was expected. Students often faced multiple challenges, one led to another, which resulted in the failing grades. Our study contributes a set of effective text analytics techniques for extracting insights from student data, providing the preliminary guidelines for an information system to detect early at risk students.

KEYWORDS:

• At risk student
• academic failure
• text analytics
• higher education institution
• unstructured data
• user-generated content

Disclosure statement

No potential conflict of interest was reported by the authors.