A data-driven perspective to the forecasting problem
Martedì 18 Maggio 2021 - ore 14.30
Michele Piana
Dipartimento di Matematica - DIMA
Università di Genova
This talk will deal with forecast verification and loss functions design as two intertwined concepts in supervised machine learning. Focusing on binary classification, we will show how this approach impact the forecasting performances of regularization networks in the case of experimental problems in different application domains