In this article we look at the apt support for Python in the 2017 edition of SQL Server
The R Services in SQL Server are now known as SQL Server Machine Learning Services; this service now allows you to make use of not just R language, but also Python. The Machine Learning Services feature in databases allows you to run scripts written in Python or R, when using SQL Server. However, if you do not have access to this service, you can also opt for downloading and installing the MS Machine Learning Server, for deploying and using editions of R and Python that do not make use of SQL Server. In the 2016 edition of SQL Server, users could only make use of R Language, but with the 2017 edition of SQL Server, users have been given access to using Python in the application as well. You can now run machine learning solutions in not just R but also Python
SQL Server developers operating in open source ecosystem, along with having access to Python ML, also make use of AI libraries, in addition to all the features introduced by Microsoft, a few of which are described below.
- Revoscalepy – This is the Python version of RevoScaleR, it makes use of parallel algorithms for the purpose of logistic and linear regressions. This also includes boosted trees, decision trees, random forests, and extensive APIs for the purpose of moving data, transforming data, etc.
- Microsoftml – This is one of those machine learning algorithms that makes use of fast decision trees and forests, extensive neural networks, along with other optimized algorithms that are put to use for logistic and linear regression.
- Using Python with T-SQL – Deploying Python has now become easier, because you can make use of the ‘sp_execute_external_script’ stored procedure. If you make use of MPI ring parallelization for streaming SQL data to Python processes, you can get access to great performance.
- Native Scoring – T – SQL consists of a PREDICT function that allows the user to score any given SQL Server instance of 2017 edition, even if you do not have R installed in your system. All that you need to do is; train the model by making use of revoscalepy and RevoSacleR algorithms that are supported. End by finally saving the model in one of the latest comact binary format.
- Package Management – CREATE EXTERNAL LIBRARY statement is now supported in T –SQL, this allows database administrators to get greater management control to the R Packages. These R packages can be shared among users by storing in databases, and roles should be used by users for controlling the private as well as shared package access.
- Performance Improvements – “sp_execute_external_script” stored procedure is now optimized for supporting execution in batch mode for data in columnstore.
- Python for SQL Server Compute Contexts – SQL Server developers as well as data scientists now have the power to remotely execute Python code from development ecosystem, for the purpose of exploring data and developing models. And also ensuring that data is not moved around.
Even the venerable and highly sophisticated SQL Server 2017 is open to flaws. Hence companies must invest in a robust tool to fix SQL Server and deal with emergency situations.
Victor Simon is a data recovery expert in DataNumen, Inc., which is the world leader in data recovery technologies, including Access corruption and sql recovery software products. For more information visit https://www.datanumen.com/