Pickle to API — Azure

So, pickle to API without writing any code. How exactly is it possible?

Using Azure’s Machine Learning Service we can do just that!

First navigate to Azure ML service from the Azure Dashboard.

You will find a page that looks like so:

Create a workspace using the following configurations for a basic and minimal set-up.

Once you’re done with the above set up tags (if required), review and create.

Navigate to ML studio after the workspace has been created.

Go to models and register your model:

Once that’s done, you will have to deploy your model. Select the model you wanna deploy

Choose the type of deployment you wanna do, for this example I’ve deployed to a web service.

Use the following configurations and deploy.

It takes about 20 minutes for the model to deploy, so make sure you have some music to keep you company!

While waiting you can check up on your model by navigating to endpoints > model.

Here, you will see that the deployment state changes frequently from unhealthy to transitioning to nothing. But patience is the key here, make sure you wait till the state is either healthy or failed till you delete it.

You can use this guide as a reference to what other deployment states you’ll encounter;

So, once your model has been successfully deployed, it should look like this:

All the best!