Flooding is a common threat worldwide. Knowing what happens in case of such events can increase the chances of human survival and property protection. The team performed data connectivity, metadata discovery, data quality inference and data exploration on the data sources to better understand the main features associated with flooding events over time. We created several machine learning models to predict flooding events, compared them and published champion and challenger models. Flooding is all about mitigating and anticipating events to help people. In addition to the ML models to predict flooding events, we developed additional models to mitigate damage and help people. We developed an optimization model to identify the optimal flows to evacuate displaced people to public shelters based on their capacity and distance. We also developed an optimization model to find the optimal routes and trips to evacuate displaced people to the identified shelters based on the number of vehicles available and their capacity. The team created a forecasting model with autocorrelation analysis to identify a set of parameter estimates associated with variables that impact the likelihood of flooding events. A what-if analysis was created to simulate the impact of flooding events while changing the values of the significant variables. Finally, a regression model on referenced spatial data was developed to estimate possible locations within hurricane paths to prepare a set of mitigation actions to help people in danger.
