http://repositorio.ufc.br/handle/riufc/21999 2026-06-17T07:17:46Z 2026-06-17T07:17:46Z Souza, Iarley Natã Lopes http://repositorio.ufc.br/handle/riufc/85887 2026-04-15T19:08:24Z 2026-01-01T00:00:00Z

Título: Avaliação do modelo TabPFN na predição de diabetes com variação no número de instâncias Autor(es): Souza, Iarley Natã Lopes Abstract: Diabetes mellitus is one of the most common chronic diseases worldwide, representing one of the greatest public health challenges of today. This condition affects the body’s ability to produce or properly use insulin, the hormone responsible for regulating blood glucose levels, contributing to the development of secondary complications and highlighting the importance of early diagnosis. With advances in machine learning techniques, these approaches have been increasingly used to support the prediction of diseases such as diabetes; however, the scarcity of clinical data, linked to data protection policies and resource limitations, still represents a significant obstacle. In this context, the TabPFN model stands out as a Transformer-based network capable of performing fast and accurate predictions even with limited amounts of data. Given this scenario, this work evaluates the performance of TabPFN in predicting cases of diabetes mellitus, analyzing its behavior under different data volumes and comparing it with well-established models in the literature. The results demonstrated that TabPFN achieved highly competitive performance, standing out from 100 training instances onward and reaching an F1-score close to 0.80 with 1,000 samples, even without the use of balancing techniques, while also identifying glucose-related variables as the most decisive for classification. It is concluded that TabPFN represents a promising alternative for diabetes prediction in data-scarce scenarios. Tipo: TCC

2026-01-01T00:00:00Z Alves, Guilherme Girão http://repositorio.ufc.br/handle/riufc/85886 2026-04-15T19:00:55Z 2026-01-01T00:00:00Z

Título: Sistema de recomendação de pares em ambientes educacionais com base nas múltiplas dimensões do perfil estudantil Autor(es): Alves, Guilherme Girão Abstract: The current educational dynamics demand methodologies that encourage student interaction and protagonism in the learning process. In this context, strategies such as Peer Instruction have gained prominence for promoting the development of cognitive and socioemotional skills. However, manually forming student pairs poses a challenge for educators, especially when aiming to consider the students’ multidimensional profiles. Existing approaches in the literature are generally limited to using these profile dimensions in isolation, which may compromise the quality of collaboration. This work proposes the development of a peer recommendation system based on the complementarity of multiple student profile dimensions, using Genetic Algorithms as an optimization technique to form complementary pairs. The solution aims to support teachers in creating collaborative interactions among students, contributing to the effectiveness of collaborative learning, particularly Peer Instruction, in the classroom. Tipo: TCC

2026-01-01T00:00:00Z Lima, Gabriel Dias de http://repositorio.ufc.br/handle/riufc/85885 2026-04-15T18:54:42Z 2026-01-01T00:00:00Z

Título: Uma comparação de desempenho entre dois algoritmos de busca tabu aplicados ao problema capacitado de localização de facilidades com cobertura parcial e fonte única Autor(es): Lima, Gabriel Dias de Abstract: In problems related to partial covering and facility location, the objective is to select a subset of facilities J′ from a set of candidate facilities J, so that at least a minimum amount of demand from a set of customers I is met. This work explores a variation of this problem that adds constraints regarding the capacity of the facilities and the requirement that customers be served by at most one facility — the single-source constraint. According to the literature review done, the most recent work exploring this same problem was developed by Mourão et al. (2024). In his study, he employed the tabu search metaheuristic combined with an integer programming solver to solve specific cases related to demand allocation. His results achieved high precision and faster execution times than those of the CPLEX solver when solving the same problem. However, the use of deterministic techniques to assist the metaheuristic may make it difficult to understand its standalone effectiveness. Therefore, in this work, the tabu tearch conceptualized by Mourão et al. (2024) was implemented in Python 3.14, and a comparative analysis between the performance of the two versions, the pure tabu search and the one aided by the CPLEX, was conducted by comparing the quality of the solutions and computational time spent. It was observed that, even without the aid of deterministic methods, the metaheuristic alone is able to f ind solutions of equivalent quality to CPLEX in 32.05% of the 78 tested instances, while its execution time decreased in 56.41% of the instances. Tipo: TCC

2026-01-01T00:00:00Z Sousa, Tiago Rodrigues http://repositorio.ufc.br/handle/riufc/85884 2026-04-15T18:48:33Z 2026-01-01T00:00:00Z

Título: Pé na Pedra: uma aplicação móvel para a equipe de trilha Pé na Pedra Autor(es): Sousa, Tiago Rodrigues Abstract: Quixadá, located in the Central Sertão region of the state of Ceará, Brazil, is internationally recognized for its inselbergs, rocky formations that create a unique landscape favorable to the practice of adventure activities. The region presents high potential for ecotourism, although many of its natural areas remain underexplored and have low visibility. In this context, this work presents the development of the Pé na Pedra application, a mobile app created to support the hiking group “Trilheiros Pé na Pedra” in sharing, organizing, and discovering ecological routes in Quixadá. The project was conducted based on the Star Life Cycle methodology, prioritizing user involvement throughout all stages of development. The results were obtained through the application of questionnaires, contextual inquiry during hikes, definition of user profiles and personas, requirements specification, prototyping, and application implementation. The usability evaluation of the prototype identified critical issues related to error prevention and recognition rather than recall, which were addressed during implementation. The final version of the application was evaluated using the System Usability Scale (SUS), achieving an average score of 90, indicating a high level of usability and user satisfaction. The results demonstrate that the application significantly contributes to supporting ecotourism practices, organizing hiking trails, and promoting lesser-known natural areas of the region. Tipo: TCC

2026-01-01T00:00:00Z