https://repositorio.ufpb.br/jspui/handle/123456789/36517 Sat, 26 Sep 2026 22:12:55 GMT 2026-09-26T22:12:55Z http://repositorio.ufpb.br:443/jspui/retrieve/163331/logo-cdn.png https://repositorio.ufpb.br/jspui/handle/123456789/36517 https://repositorio.ufpb.br/jspui/handle/123456789/39216 Título: Previsão de rebaixamento no campeonato brasileiro Série A: uma comparação de modelos de machine learning com validação walk-forward Autor(es): Barroso, Leonardo Feitosa Orientador: Ramalho, Hilton Martins de Brito Abstract: Relegation to Série B is one of the greatest sporting and financial losses a Brazilian football club can suffer, directly affecting broadcasting rights revenues, sponsorships, and squad market value. The objective of this study is to develop and compare Machine Learning models capable of predicting, before the start of the season, which clubs face the highest relegation risk in the Brazilian Série A championship. Four binary classification algorithms were evaluated — Logistic Regression, Random Forest, XGBoost, and LightGBM — over a set of 15 features combining squad financial data, collected from the Transfermarkt portal, with sliding windows of performance from the previous three and five seasons. Validation used the walk-forward scheme, which respects the chronological order of the data, and hyperparameter optimization was performed with RandomizedSearchCV combined with TimeSeriesSplit. On the independent test set (2023–2024), LightGBM achieved the best AUC-ROC (0.877), while Logistic Regression showed the highest temporal stability in walk-forward validation (mean 0.794 ± 0.058). The forecast for the 2025 season pointed to Juventude, Sport Recife, Vitória, and Mirassol as the clubs at highest risk; retrospective validation against the official league results confirmed two of the four predicted relegations and an AUC-ROC of 0.922 for the season, with all four actually relegated clubs ranked among the seven highest predicted risks. The study contributes a reproducible pipeline for sports prediction with temporal data, empirically demonstrates the inadequacy of accuracy as a sole metric in imbalanced classification problems, and delivers the final model as an openly accessible web application. Editor: Universidade Federal da Paraíba Tipo: TCC Mon, 10 Aug 2026 00:00:00 GMT https://repositorio.ufpb.br/jspui/handle/123456789/39216 2026-08-10T00:00:00Z https://repositorio.ufpb.br/jspui/handle/123456789/38872 Título: RecomendaAí: do pipeline de dados à recomendação personalizada em aplicação web Autor(es): Carlos, Lorena Ximenes Orientador: Viana, Jorge Henrique Norões Abstract: The RecomendaAí is a web application for movie and TV series recommendation developed in the context of Applied Data Science, focusing on mitigating choice overload in extensive audiovisual catalogs. The solution uses real data from The Movie Database (TMDb) API, consolidated into a processed catalog of 18,128 titles, to generate recommendations within an integrated environment that includes catalog exploration, authentication, personaliza- tion, and user interaction tracking. The methodology combines content-based filtering through TF-IDF (Term Frequency–Inverse Document Frequency) vectorization and cosine similarity, with optional semantic search via embeddings and hybrid reranking strategies based on behavioral signals and usage context. The application organizes the user experi- ence into the Search, Discover, and For You tabs, enabling recommendations from known titles, natural-language queries, and continuous personalization from interaction history. This work demonstrates the integration of Data Science techniques and web development in building a functional, interpretable, and usable recommendation system capable of reducing the user’s search space in large catalogs. Editor: Universidade Federal da Paraíba Tipo: TCC Tue, 11 Aug 2026 00:00:00 GMT https://repositorio.ufpb.br/jspui/handle/123456789/38872 2026-08-11T00:00:00Z https://repositorio.ufpb.br/jspui/handle/123456789/38870 Título: Dos dados ao alerta: utilizando machine learning para prever e explicar a gravidade de acidentes em rodovias federais brasileiras Autor(es): Reis Sousa, Elaine Regina Orientador: Viana, Jorge Henrique Norões Abstract: Traffic accidents on federal highways represent a complex challenge for public safety and infrastructure management in Brazil. This study aims to develop an interactive web application based on Machine Learning to estimate the probability that a person involved in an accident will experience a severe outcome (serious injury or death) and to explain the factors associated with the prediction. The methodology was based on training a predictive model using the XGBoost algorithm, fed with historical accident data and information on the vehicle fleet. To ensure system transparency and avoid the opacity inherent in complex models (the “black-box” effect), the SHAP (SHapley Additive exPlanations) library was integrated to provide local explainability for the predictions. The solution was developed using a modular architecture, separating the web interface, built with Streamlit, from the services responsible for data loading, input processing, prediction generation, and result interpretation. The results demonstrate that the dashboard allows users to enter physical, temporal, and environmental characteristics of a road segment and, in real time, estimate the probability that a person involved in an accident will sustain a serious injury or die, while also displaying a graph detailing the contribution of each variable to the prediction. It is concluded that the tool functions as a predictive and interpretable decision-support system, with practical potential to assist agencies such as the Brazilian Federal Highway Police (PRF) and the National Department of Transport Infrastructure (DNIT) in the preventive allocation of enforcement resources and the prioritization of improvements to the road network. Editor: Universidade Federal da Paraíba Tipo: TCC Tue, 11 Aug 2026 00:00:00 GMT https://repositorio.ufpb.br/jspui/handle/123456789/38870 2026-08-11T00:00:00Z https://repositorio.ufpb.br/jspui/handle/123456789/38163 Título: Estudo sobre a evolução da eficiência das Odds no mercado de apostas de futebol Autor(es): Flôr, Matheus Santos de Oliveira Orientador: Viana, Jorge Henrique Norões Abstract: This study aims to analyze and compare the degree of informational efficiency in football betting markets for the English Premier League and the Brazilian Championship (Série A), with the purpose of identifying the presence, magnitude, and possible causes of pricing inefficiencies in the odds. The methodology was based on building a database through automated extraction (web scraping) and the integration of weather variables. The predictive modeling, focused on binary outcome classification, compared five machine learning algorithms, optimized via stochastic search and validated by the intertemporal walk-forward method to prevent data leakage. The prediction testing occurred in a simulated financial backtest environment. Statistical results demonstrated the superiority of Logistic Regression with L2 regularization (60.17% accuracy and 0.6202 AUC in the test set), which outperformed complex architectures based on decision trees. In the financial application guided by Expected Value (+EV) thresholds, a market asymmetry was observed: the algorithm achieved a positive Return on Investment (ROI) of 0.78% in the Brazilian Championship, but recorded a loss of -38.35% in the Premier League. It is concluded, further supported by Monte Carlo simulations, that the English market validates the Efficient Market Hypothesis in its semi-strong form due to its high liquidity, nullifying basic statistical advantages, whereas the Brazilian scenario, in its consolidation phase, still presents marginal inefficiencies that can be quantitatively exploited. Editor: Universidade Federal da Paraíba Tipo: TCC Thu, 09 Apr 2026 00:00:00 GMT https://repositorio.ufpb.br/jspui/handle/123456789/38163 2026-04-09T00:00:00Z