I normally work with most of the Microsoft tools for the aim of doing machine learning. I came up with a roadmap that shows what option for machine learning we have with Microsoft tools in Data Platform. this roadmap is not completed yet, I need to update it one later with some tools such azure data bricks, data science virtual machine and so forth. But, in this post, I try to provide an overview what options we have regarding working with MS ML tools.
Writing R or Python in some Microsoft data Tools
There is a possibility to write your own R or Python codes inside some of the Microsoft data platform tools.
Azure ML
Azure mL studio is a drag and drops environment that helps users to create an ML model using some predefined model to create a machine learning models and deploy it as an API. However, besides 25 pre-defined algorithms that already exists in Azure ML Studio, there is a possibility to write Python or R codes inside the Azure ML studio. for more information click here.
SQL Server 2016 and 2017
Moreover, there is a possibility to embed R or Python codes in SQL Server 2017 (ML services) or only R in SQL Server 2016 (R services). I have some blog posts on this that show the process you see from post1, post2
Power BI and R
there is a possibility to run R codes in Power BI with the aim of visualization in the report area, create the custom visual using R and JSON, or using R inside the Power Query. the detail of how we can do that has been explained in posts
Azure ML work Bench with Python
Azure ML workbench as an on-premises tool able to help people to easily deploy their codes. There is a possibility to write Python codes and deploy the code. in near future, I will write some posts on how to write Python codes inside Azure ML workbench.
Azure Data Lake write R and Python
there is also to write R or Python codes inside the U-SQL language in Azure data Lake.
Use the created ML model (API) in other Applications
As you know there is a possibility to create API from our Machine Learning model. This model can be consumed in Excel (add-on), Stream Analytics, and Power BI. The same for cognitive services API that can be consumed in Power BI, .Net. and Microsoft flow.
Access to data resources from R studio for data modeling and exploration
For most of the data science application, we need normally to create the code and test it in R studio, then use the codes in other application. For this aim, there is a possibility to load the data from SQL Server 2016 and 2017 into R studio using “RODBC” package (see post). Also, there is a possibility to get the data from Azure Data Lake and do analysis on them (see post). Finally, we able to fetch the model and datasets from Azure ML or upload data into Azure ML environment via R Studio (see posts).
