PREDICTING SEVERE RAINFALL IN SÃO PAULO WITH MACHINE LEARNING MODELS

Climate change has made the prediction of intense rainfall increasingly challenging, impacting urban management. The objective of this study was to evaluate the application of machine learning methods in forecasting intense rainfall in São Paulo. The methodology included the application of PCA combined with K-means for data segmentation and the use of models such as Logistic Regression, Random Forest, SVM, and Gradient Boosting. The results showed that Random Forest stood out with high values of precision, recall, F1-Score, and accuracy, indicating its robustness and superior generalization capability. In particular, Random Forest achieved a recall and F1-Score of 0.87 and an AUC of 0.81, demonstrating its effectiveness in discriminating between rainfall classes. The integration of machine learning techniques proved effective in forecasting intense rainfall, which can contribute to risk mitigation strategies.