What is online learning in time series forecasting systems?

Online learning updates models continuously as new time series data arrives.

Online learning systems update model parameters incrementally rather than retraining from scratch. This is essential for real-time forecasting systems where data arrives continuously. Techniques include stochastic gradient updates, streaming ARIMA variants, and incremental deep learning models.