Aplicação de Regressão Logistica em Manutenção de Vias de Rolamento Urbanas

Urban road infrastructure plays a crucial role in Brazil’s socioeconomic development, as it is essential for the transportation of goods and passengers. However, it faces significant challenges, such as pavement deterioration due to heavy traffic and a lack of preventive maintenance. This study aims to assess the use of logistic regression as a tool for predicting fatigue cracks in urban pavements. Using a fictional database with variables such as absolute slope, average daily traffic volume (ADT), and percentage of heavy trucks, the analysis was performed using the JASP software. The results showed that the percentage of trucks has a significant impact on pavement deterioration, followed by road incline and traffic volume. Although the model demonstrated an accuracy of over 70%, specificity was relatively low (around 50%), indicating the need for refinement to better predict segments without faults. The use of a fictional database, while useful, limited the representativeness of real roads. Future studies should prioritize the collection of real data, increased sample size, and experimentation with alternative techniques such as artificial neural networks and decision trees to improve model accuracy and optimize urban pavement management.