Deploy the analytics as a service that can automatically generate the endpoints with which the external applications can interact with the model. When a machine learning code is written and published to a registry, it can provide the following endpoints automatically, so that the client applications can send events such as train, test, and predict to the respective endpoint.
The following are a set of endpoints for a machine learning model:
- model <x>/train: For training the model so that the external system can send a train event
- model <x>/test: For testing the model so that the external system can send a test event
- model <x> /validate: For validating the model so that the external system can send a validate event
- model <x>/predict: ...