Developing our first MaaS with MLflow

In this section, we will (re)develop the digital twin of our wind turbine seen in Chapter 14 using real data that you can find in the official repository of this book at https://github.com/PacktPublishing/Hands-On-Industrial-Internet-of-ThingsSecond-Edition. Then, we will deploy it using a popular ML operations (MLOps) platform called MLflow (https://mlflow.org). MLflow, based on Python, was born as an experiment tracking platform and extends its capabilities to model storage and serving.

Starting MLflow

To start MLflow, from the command console, install and start the service as shown here:

$ pip install mlflow
$ mlflow server --host 127.0.0.1 --port 8080

Now, you can connect to http://localhost:8080 to see the web interface.

However, using MLflow as a standalone service is very limiting because its architecture is a bit more complex. MLflow, in fact, requires some services such as Simple Storage Service (S3) and Azure SQL Database...