Over the first decades of the 2000s, competition has brought the best public clouds close enough both in terms of the products they offer and their prices. While I was writing this book, it was really hard to predict that a single public cloud would prove to be absolutely better than the others and overtake the market.

Clouds provided by Amazon (AWS), Google (GCP), and Microsoft (Azure) are all very good. Even though picking any of these would not be a mistake, this section is about to highlight the strengths of Azure while discussing why Machine Learning Studio is a good product to get yourself started with clouds.

First of all, Azure is provided by Microsoft, which means integration with a lengthy list of other products they provide. These integrated services are likely to play a very important role for medium-sized and large enterprises, usually as Business Intelligence (BI) tools.

If your business relies on Microsoft products, consider using Azure cloud. There are loads of compatibilities and synchronization tools that might help you a lot. Taking Microsoft's popularity into account, its ease of use may be a great advantage.

Being Windows-synchronizable might be a unique characteristic, but it's not the only one; compliance is another strong point. Updating data to a place where it shouldn't be can turn into a Godzilla-sized problem with huge consequences. Azure comes with many compliance certificates and attestations so that companies are more inclined to see it as a trustworthy data and application holder.

Counter to intuition, Azure also supports lots of open source tools besides the paid ones provided by Microsoft. You can launch a (virtual) machine set with CentOS as easy you can launch one based on Windows 10. Azure is not restricted to Windows' products at all.

All of this, combined with some other stuff I didn't mention, makes Microsoft Azure a very competitive cloud in general. Nonetheless, the current chapter won't discuss a lot of general aspects. It rather focuses on a single service called Machine Learning Studio.

Machine Learning Studio would be better labeled as a PaaS. Through a well-designed, intuitive platform, you can perform data science as easily as drawing diagrams. Machine Learning Studio works in a drag and drop system. With little technical knowledge and some theoretical understanding, preset operations such as load CSV, clean data, fit model, and export results can be chained together to form a whole data science work.

If you happen to know R, greater flexibility can be achieved because you can drag and drop and run R scripts there. Mastering Python is also good. Python scripts are also supported—iterating R and Python scripts is yet another option. The drag and drop interface is not only easy to use, it's responsible for the Machine Learning Studio's cool diagram layout.

Such a layout is great for a series of reasons. First, users can visualize how each stage is interacting to form the whole. Secondly, it favors organization and inspection, while allowing corrections and improvements to be deployed as fast as anything is spotted or imagined. Thirdly, this form is convenient for teamwork.

Colleagues can easily understand what has been done, take from it, and even promote discussions. Machine Learning Studio is a whiteboard where you and your team can design data science solutions with a pen and implement them as soon as you've finished drawing, by simply clicking a button.

Machine Learning Studio does support teamwork by allowing different users to work and interact across the same project within the same platform. I would not recommend it if a very sophisticated level of flexibility is required; otherwise, that is a great product both for lone workers new to data science and for experienced parties with a wide range of skills. The next section is meant to guide newcomers through the first steps into Azure.