This study presents a pioneering approach to predict dengue outbreaks across Brazilian municipalities by integrating epidemiological, climatic, and socioeconomic data through the full Knowledge Discovery in Databases (KDD) process. We analyzed a comprehensive dataset from TABNET, INMET, and IBGE, identifying key predictors such as rainfall, temperature, and population density. Advanced machine learning models, including Random Forest and SARIMAX, were developed, demonstrating high accuracy in forecasting epidemic peaks. The results underscore the potential of datadriven strategies to enhance epidemiological surveillance and support proactive public health interventions, offering a robust, scalable framework for effectively mitigating the impact of dengue in Brazil.