Análise da trajetória acadêmica estudantil utilizando mineração de processos com ênfase na evasão do curso de Ciência da Computação da UFC em Quixadá

Abstract

Higher education institutions face several challenges, among which student dropout is one of the most complex and concerning. Course abandonment by students leads to individual, social, and institutional losses. In this context, the application of Process Mining techniques emerges as a promising analytical tool to understand the academic trajectories that lead to dropout. This study proposes the modeling of the curriculum structure of the Computer Science program at the Federal University of Ceará (UFC), Campus of Quixadá, using event logs extracted from students’ academic records. The dataset includes information about the courses taken, their status (enrolled, approved, failed, failed due to absence, withdrawal, canceled, or full withdrawal), anonymized student IDs, and academic terms. The curriculum structure is formalized through Petri nets, enabling the representation of academic progression and the simulation of different trajectories. Subsequently, conformance checking techniques, such as token replay and alignments, are applied using the PM4PY library to assess the alignment between real academic paths and the ideal model, identifying deviations, bottlenecks, and retention patterns. The analysis revealed specific courses and semesters with high rates of failure and dropout, highlighting critical points that affect student retention. The results indicate that curriculum modeling based on process mining can support academic decision-making, contributing to reducing dropout rates and improving institutional policies for student support and retention.