Banca de QUALIFICAÇÃO: MARCELO DE ANDRADE LIMA MAIA

Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
STUDENT : MARCELO DE ANDRADE LIMA MAIA
DATE: 13/01/2026
TIME: 10:00
LOCAL: VIDEOCONFERÊNCIA
TITLE:

PREDICTIVE MODEL OF STUDENT DROPOUT IN HIGHER EDUCATION: APPLICATION OF MACHINE LEARNING AT THE FEDERAL UNIVERSITY OF RIO GRANDE DO NORTE

KEY WORDS:

Higher Education Dropout. Predictive Models. Machine Learning. Student Retention Management. Design Science Research.

PAGES: 60
BIG AREA: Ciências Sociais Aplicadas
AREA: Administração
SUMMARY:

Dropout is a critical challenge in Brazilian higher education system, with cumulative rates approaching 60% nationwide (INEP, 2024), creating waste of institutional resources and human capital. Research on predictive models for attrition remains limited, especially when it comes to recent machine-learning approaches. This dissertation aims to develop a predictive model of student attrition using Machine Learning techniques as a decision-support tool for retention policies. The study is applied in nature and follows Design Science Research as its methodological strategy for building and evaluating the model, drawing on Vincent Tinto’s Student Integration Theory and Bean and Metzner’s model for nontraditional students as Kernel Theories, which emphasize academic and social integration, environmental factors, and students’ living conditions as drivers of attrition. The empirical basis consists of administrative and academic data from the Federal University of Rio Grande do Norte (UFRN) between 2015 and 2024, extracted from the SIG ecosystem. The developed artifact includes the structuring and validation of a predictive model based on Machine Learning algorithms under metrics suited to imbalanced data, focused on early identification of high-risk students; the identification of key predictors of dropout in light of Tinto and Bean–Metzner; and the definition of guidelines and requirements for implementing the tool in managerial monitoring dashboards integrated into UFRN’s institutional systems.

COMMITTEE MEMBERS:
Externo à Instituição - JORGE HENRIQUE NOROES VIANA - UFPB
Presidente - 1510488 - LUCIANO MENEZES BEZERRA SAMPAIO
Interna - 1894891 - RAQUEL MENEZES BEZERRA SAMPAIO