Microsoft Azure Machine Learning Studio Review

Easy to deploy, drag and drop makes it easy to test various algorithms


What is our primary use case?

The first time that I used this tool was in a project related to bike usage in the city of Boston. This project was part of a course that I concluded some months ago. In this project I used components to read data, for exploratory analysis, for steps of data munging, to split data, select hyperparameters, and some machine learning algorithms. In some steps I needed to insert R modules to apply some data transformation.

The target of this exercise was to predict bike usage in a day.

How has it helped my organization?

With this tool we could have all benefits of a cloud environment, such as scalability and access to machine-learning applications. These features are very important when you have large datasets and critical applications.

What is most valuable?

  • It is very easy to test different kinds of machine-learning algorithms with different parameters. You choose the algorithm, drag and drop to the workspace, and plug the dataset into this component.
  • When you import the dataset you can see the data distribution easily with graphics and statistical measures.
  • Easy to deploy and provide the project like a service.

What needs improvement?

For my project/exercise, this tools was perfect. I would like to see modules to handle Deep Learning frameworks.

For how long have I used the solution?

Less than one year.

What do I think about the stability of the solution?

No issues with stability.

What do I think about the scalability of the solution?

No issues with scalability.

How is customer service and technical support?

I didn’t need to use the support, but this tool has great documentation.

Which solutions did we use previously?

Nowadays I use Python (Anaconda and Jupyter Notebook) and R (RStudio) to create my solutions and machine-learning models.

How was the initial setup?

It was very simple and straightforward. It is really simple to start building a project.

What's my experience with pricing, setup cost, and licensing?

There are two kinds of licenses, Free and Standard.

Free

  • 100 modules per experiment.
  • 1 hour per experiment.
  • 10GB storage space.
  • Single Node Execution/Performance.

Standard – $9.99/seat/month (probably a data scientist)

  • $1 per Studio Experimentation Hour. You will pay according to the number of hours your experiments run.
  • Unlimited modules per experiment.
  • Up to seven days per experiment, 24 hours per module.
  • Unlimited BYO storage space.
  • On-premises SQL data processing.
  • Multiple Nodes Execution/Performance.
  • Production Web API.
  • SLA.

What other advice do I have?

You will be able to create your machine-learning project and extract insights from it just by dragging and dropping components and adjusting some parameters. This tool is very user-friendly, so without a lot of programming skills you can build machine-learning projects. 

If you need more control over machine-learning modules you will need to add R or Python modules to create a customized machine-learning model.

Disclosure: I am a real user, and this review is based on my own experience and opinions.
Add a Comment
Guest
Sign Up with Email