The use of machine learning to address floods

This study evaluates the effectiveness of applying artificial neural networks as a decision-support tool during natural disasters, focusing on the floods that occurred in Pelotas, Rio Grande do Sul, in 2024. Using a machine learning approach, the study analyzes whether the decisions made by the municipal government were conservative or flexible by comparing water level forecasts with the implemented actions. Based on time series data from the São Gonçalo Channel, the results indicate that the model's predictions generally aligned with public policies, though they slightly overestimated water levels. The findings suggest that integrating predictive models based on artificial intelligence can enhance the effectiveness of public policies and risk management during climate emergencies. This study contributes to the literature by demonstrating the potential of data science techniques for public managers, aiding in damage mitigation and the formulation of proactive civil protection strategies.