In the process of building a machine learning model, an interface needs to be designed on how to interact with the model. One of the approaches discussed here is an event-driven machine learning model that can be triggered based on different kinds of event data and where the necessary action can be triggered for predict, train, retrain, and validate.
In a event-driven model, the event definition will determine the type of action that needs to be triggered and also provides the flexibility on how to fetch the data. Typically, for real time model predictions, it needs to provide the predictions in real-time in continuous mode, for which the model can receive a stream of events from a messaging channel, ...
