Getting Started with Azure ML Studio

Microsoft has released Azure machine learning studio which is a browser based integrated development environment for constructing and operationalizing machine learning workflow on azure.  You can find more details about the tool here – https://studio.azureml.net/.

We at springML love programming models in R and Python.  These programming languages along with the myriad of supporting libraries provide a powerful platform to run the end to end process of model creation and deployment.  However these languages have a steep learning curve which is what the AzureML Studio solves pretty elegantly.

AzureML Studio is simple and easy to use and since it’s browser based there’s nothing to install.  You can get started within a few minutes.  I was able to quickly sign up using my hotmail account (who knew my old Hotmail account would come in handy someday ;)) .  I was then able to browse and look at various existing tutorials and samples and import code in to my Studio with the click of a button.  The GUI itself is pretty slick.  You can drag and drop various components on to the canvas and set properties as appropriate.  You can also execute the models and visualize the results.  AzureML Studio also provides an easy way to include custom R or Python code in case geeks like me miss it.

There are plenty of tutorials and samples available that you can import by the click of a button.  This makes the learning process easy and one can ramp up pretty quickly especially if that person has prior knowledge of machine learning models.

azure1

AzureML support various out of the box components to create a model.  Here’s a screenshot showing the various capabilities.

azure2

One of the coolest features I found on AzureML was the ability to publish a model as a web service.  This is the most seamless process I have seen.  There are not a ton of options to deploy an R model as a web service (you may want to do this in order to consume it from a mobile device, for example) and I found the solution from AzureML incredibly elegant.

They have a marketplace and I am looking forward to contributing a few models soon.