Selecting optimal architecture for time series forecasting remains challenging. Here, the authors present an automated framework that optimizes the composition of hybrid models via multi-objective ...