Machine Learning with Python

Support for the widely used R statistics language was introduced in SQL Server 2016. Using the predefined stored sp_execute_external_script procedure, R functions can be encapsulated in their own procedures, provided with the desired arguments and result transformations, and then used in SQL databases.

SQL Server 2017 natively supports another scripting language that is widely used, especially in the areas of IoT and machine learning: Python. This capability allows SQL projects to bring together the best of both worlds: robust and fast relational data management with SQL and an almost infinite stock of methods and libraries that have already been developed in Python for machine learning. Complemented by the precise, high-performance statistical functions of R, it provides an ideal platform for big data, IoT, and machine learning applications. According to a 2017 software poll [9], SQL Server 2017 covers the three most widely used languages for these applications.

R Services for SQL has therefore also become machine learning Services. However, you do not necessarily have to install support for both R and Python – you can still install only R if your projects do not use Python.

Conclusions

SQL Server 2017 is not just a great technological milestone; it also further cements the direction of development taken in the previous version. Even if your organization's product life cycles do not require the next version of the database platform for the time being, you might want to start working with SQL Server 2017 now. Linux and container support have the potential to enable a change of strategy on the infrastructure side. The new machine learning services, support for graph data, and the ability to operate Power BI locally can give a new impetus to your internal data processing.