Understanding Azure Machine Learning Studio

What is Azure Machine Learning Studio?

It is an online IDE which provides a simple and easy drag-and-drop approach to build, test, and deploy predictive analytics solutions on data. With Azure Machine Learning Studio, you can publish models as web series, which may be used by other applications or tools like Excel.

Azure Machine Learning Studio provides an interactive, visual workspace where it is easy to build, test, and iterate on a predictive analysis model. It follows a drag-and-drop approach where datasets and models can be put on an interactive canvas connecting these together to form an experiment which runs as a solution on Machine Learning Studio. We can edit the experiment at any time and save a copy of the required ones and run our experiment to get the desired task completed.

One of the striking features of Azure Machine Learning Studio is that we can publish the trained predictive models as a web service.

All in all, Azure Machine Learning Studio is a platform where cloud resources, data science, predictive analysis, and the data meet together.

Hey, before you read ahead, would you like to learn the basics of Machine Learning? I’ve written an article on First Guide to Machine Learning; consider reading it.

Get started with Azure Machine Learning Studio

The first step, no doubt, is to visit here. By default, you will land to the Homepage. On the top left, you will find the hamburger menu with three categories (a) Cortana Intelligence (b) Azure Machine Learning Studio (c) Gallery. Let us discuss each one of them briefly.

Azure Machine Learning Studio
Figure Menu

Cortana Intelligence

It is a fully managed space that provides various end-to-end solutions to transform data into intelligent action. This platform usually talks about the productive, hybrid, intelligent, and trusted cloud platform, i.e., Azure that can be used to turn your ideas into solution faster. It has ready-to-go solutions to start building right from the moment you think of an idea.

Azure Machine Learning Studio
Figure End-to-End solution available in Cortana Intelligence


It is actually an Azure AI gallery which is a community of data scientists and developers where they share their solution with other members of the community.

Azure Machine Learning Studio
Figure Solution shared by community members

Azure Machine Learning Studio

This category has two options Home, the default page you landed and Studio where we are going to spend rest of our time. Click on Studio and you will be directed to Sign-In page. You may use any Microsoft account, or work account or school account.

Azure Machine Learning Studio
Figure Click on the Studio Link
Azure Machine Learning Studio
Figure Sign in with Microsoft account or work or school account
Azure Machine Learning Studio
Figure Type your password here

You can also directly Sign-In from the Home page and navigate to Studio from the menu.

Azure Machine Learning Studio 

On successful sign-in, you will be directed to the Azure Machine Learning Studio environment. On your left, you will find tabs. Let us discuss each one of them.

Azure Machine Learning Studio
Figure Panel on the left side


It is one single entity for a particular project that might contain datasets, notebooks, trained models and other resources related to a single project.


An experiment might contain models and logic that we have created. You can RUN, EDIT, SAVE and DELETE experiment.

Web Services

All the web services that you have deployed can be found here.


These are actually Jupyter Notebooks. A simple, interactive and easy to read python environment. All jupyter notebooks saved can found here.


Here you can see all the data sets that you have uploaded in Machine Learning studio.

Training Model

An experiment consists of various models. If you would like to save your train mode, this is the place you come to have a look at them.


This section helps you configure your account and other resources.

Components of an Experiment

An experiment is a collection of datasets and models which can be connected with each other to get a predictive analytic model. All the data sets and models are places in experiment canvas. Following are the characteristics of a valid experiment.

  1. Every experiment should have at least one data set and one module.
  2. Dataset may be connected to one or more module.
  3. Modules may be connected to other modules or datasets.
  4. All input port of modules must be feed with some data.
  5. Every Module has some mandatory parameters that must be set.


Datasets can be uploaded in Azure Machine Learning Studio from various source that may be used in an experiment for training the model. Azure Machine Learning Studio provides some pre-uploaded datasets that can be used in the experiment. For example

  1. Automobile Price Data (Raw)
  2. Breast Cancer Data
  3. Forest Fires Data

Datasets may be uploaded from local storage, from online resources, or from SQL Server Database.


Algorithms in Azure Machine Learning Studio are packed in a bundle called Module. Azure Machine Learning Studio has many different tasks like training, evaluation, cleaning of data. Each module comes with a set of properties that can be configured accordingly. The properties include parameters, that can be modified to tune our model.


So now, we are done with understanding the ecosystem of Azure of Machine Learning and you are good to go ahead make Your First Experiment in Azure ML Studio.