Support for time series forecasting

[corridordigital]

Problem: time series forecasting quality depends a lot on the ML method, and it's often hard to decide or very time-consuming

Solution: a feature, in addition to regression & classification, that makes it easy to compare the most common time series forecasting models (Prophet, AutoARIMA, NeuralProphet, etc.)

Alternatives: there is no alternative currently available. GluonTS and sktime provide a more or less unified API for time series forecasting models