Applied Forecasting: From Linear Models to Data Mining and Deep Learning

Catalog description

This class guides you through each step of designing and implementing a real-world forecasting project. Its objective is to give you a working knowledge of the statistical and computational tools that are commonly involved in such exercises. We will examine variable selection, estimation, the role of hyperparameters, combination methods for forecasting models, and how to evaluate the resulting forecasts from both a statistical and a user perspective. We will emphasize the implementation and assessment of forecasting methods, including a philosophical and statistical foundation. We will work with both classical and modern methods, e.g. logistic regression and decision trees for classification tasks like credit risk assessment; or linear and nonlinear prediction models like ARx and feedforward neural networks for high and low frequency time series analysis. Throughout the semester, you will apply your knowledge in a hands-on forecasting project in areas such as business cycles, electoral processes, financial markets, or even sport events. This practical experience will solidify your understanding, prepare you for real-world challenges and elevate your forecasting skills.