Machine learning models deployment examples

Machine Learning models are often best served ... as an RESTful API to be used by an external user. A little bit of web programming (starting a Microservice for example) is needed to wrap around your core ML model. We show examples of such ML API and how to create them in this section.

A linear regression example with US Housing data

Flask and Gunicorn are used to run a simple HTTP server/interface. User can start the server, which functions as the model endpoint API, and then execute another Python script to send test data and request predictions from the pre-trained model.

A text generating app with recurrent neural network

A HTTP web app, complete with WTForms, CSS and aided by Jinja templating, is served to the user using Flask. Upon submitting the request via this web form, text is generated based on a pre-trained recurrent neural network (RNN) with word embedding.

Uh oh!