Fig. 3

Representation of the four machine learning paradigms in a batch and stream data processing system. The continuous and unlimited feature of data enables incremental learning where the model learns from incoming data. In batch processing, model training is only performed on stored historical databases