Forecasting Accurately in Space and Time (FAST): Approach and Method

This memo details the methods that the Université de Montréal team employed in the Fast Forward competition hosted by IDEaS. While the FAST project was headquartered at Université de Montréal, its members at UT-Dallas and West Virginia University led the predictive modeling. Separate teams (led by Brandt and D'Orazio respectively) developed the country and grid month forecasts, and a third (Margulies, McLauchlin and Patterson) developed our machine-generated explanations. All our work is open-source and replicable, with forecasts publicly available on Github.