Machine learning for nowcasting. The ensemble methods applied to decision trees case
Let us consider questions like: What is the current state of output? What will be the evolution of prices in the short future? Are they increasing or decreasing? Is it thundering right now and what is the situation off the sea? These questions which at a first glance seem unrelated have one thing in common; they all challenge researchers to make very quick decisions about the conditions around them, to nowcast, as it is commonly called. Every airport needs a nowcasting mechanism for the weather; likewise, every Central Bank should have a nowcasting system for economic aggregates. In this paper, new paths of answering these very simple economic questions are examined. Benefiting from the vast amount of economic data that is disposable, a family of Machine Learning techniques, Ensemble Models, are applied in the process of Nowcasting one of the most important Macroeconomic Aggregates: quarterly GDP.