Autor(es)

Ticona, Julio J. ; Nonato, Luis Gustavo ; Silva, Claudio T. ; Gomez-Nieto, Erick

• Tipo de documento: Artigo Científico
• Fonte: COMPUTERS & GRAPHICS-UK; v. 132, p. 11-pg., 2025-11-01.

Resumo

Maintaining low dropout rates remains a fundamental priority for higher education institutions. Each year, numerous students depart for various reasons, including socioeconomic challenges, academic difficulties, and social issues. For the offices tasked with monitoring enrollment and dropout trends, it is crucial to obtain a comprehensive and timely understanding of these dynamics. Regrettably, existing tools often fall short in providing an effective and straightforward means to explore and identify the key factors contributing to student dropout, thus hindering agile decision-making processes. In response to this challenge, we introduce a novel tool designed to enhance student analysis, facilitate the early detection of potential dropouts, and recommend viable strategies to mitigate attrition in higher education. This tool, named as SDR-Explorer, comprises multiple linked views that empower analysts to (i) visually monitor students' academic performance over multiple semesters, (ii) interactively examine student features to uncover patterns and clusters, (iii) predict potential dropouts for upcoming periods, and (iv) propose actionable actions over specific student characteristics to reduce dropout rates. Furthermore, the system incorporates a textual assistant that enhances the user experience by assisting in the selection, filtering, summarization, and narrative presentation of proposed changes in natural language. This feature significantly contributes to a more efficient and enjoyable analytical process. Finally, we present two usage scenarios derived from real data collected at a university, alongside a user evaluation designed to assess the usability of our system in terms of accuracy and the time required to complete analytical tasks. (AU)

• Processo FAPESP: 22/09091-8 - Criminalidade, insegurança e legitimidade: uma abordagem transdisciplinar
• Beneficiário: Luis Gustavo Nonato
• Modalidade de apoio: Auxílio à Pesquisa - Programa eScience e Data Science - Temático