10.1Introduction to Microsoft Azure Machine Learning Studio
Azure Machine Learning Studio
Now that you've learned how to prepare data for analysis and test for all assumptions (linearity, normality, homoskedasticity) before analyzing the data; and now that you've learned how to create a linear regression-based predication calculator "from scratch" in Microsoft Excel, let's learn how to use a more powerful tool that will automate some of the tedious tasks (e.g. creating dummy codes) and provide a richer selection of statistical algorithms that may greatly improve our predictive capabilities.
Microsoft's Azure Machine Learning (ML) Studio is a cloud-based tool that will support all of the primary requirements of predictive modeling with a large suite of capabilities. However, it is not the only option. Amazon Web Services (AWS) has their own cloud-based ML Studio tool. SPSS, SAS, and other statistics tools can perform the same analyses. But what sets apart the Azure and AWS is their accompanying suite of cloud server tools for database, app, and website deployment that makes it very easy to build a predictive model and integrate it into an existing software application for immediate use. SPSS, SAS, and similar products were originally developed for "offline" predictive modeling. Therefore, although those tools are constantly improving, they don't have the same level of easy integration options for deployment.
We have chosen to teach Azure ML Studio rather than AWS simply because Microsoft offers a free tier for nearly all of their cloud services making it easy to learn and practice your model deployment. However, note that AWS ML Studio is also a very good option with very similar features and is actually somewhat less expensive to scale and run in production. That being said, by learning Azure ML Studio, you will be able to migrate easily to AWS ML Studio if needed.
Another advantage of cloud-based tools like Azure (and AWS) ML Studio is the smooth visual, point-and-click interface that puts predictive modeling power in the hands of those without PhDs in statistics and computer science. The figure below visualizes the most common view:
To begin learning and using Azure ML Studio, sign up here (https://studio.azureml.net/) by clicking the "Sign up here" link and selecting the "Free Workspace"
Finally, you may want to bookmark a very useful site: Azure ML Studio Reference. This site will give you greater detail on every single "pill" that you are about to learn and all of the others we won't have time to show you.