Machine Learning With Go

Testing the Docker images locally

Before pushing our Docker-ized modeling processes to any servers, it's wise to test them locally to ensure that we are seeing the behavior that we expect. Then, once we are satisfied with that behavior, we can rest assured that these Docker images will run exactly the same on any other host that is running Docker. This assurance makes the use of Docker images a significant contributor to maintaining reproducibility with our deploys.

Let's suppose that we have our training data and some input attribute files in the following directories:

$ lsattributes model training$ ls model$ ls attributes 1.json 2.json 3.json$ ls training diabetes.csv

We can run our Docker images as software containers locally using the ...