Today I presented a talk at TechEd North America 2011, titled Enterprise Data Mining with SQL Server. I am blogging today to talk about the outline of that presentation, and provide a link to download the demo code. Also, the presentation is available on Channel 9, and I provide that link in this blog post.
The premise of the presentation is that the word Enterprise separates the SQL Server Data Mining technology from many data mining applications. Many other competing commercial or open-source solutions are for desktops. While this technology can be used on a single computer (see my description of possible configurations), the word Enterprise implies that data mining can help leverage information for large organizations. People who implement this type of solution can become leaders of leaders of leaders, true even if they do not directly manage another person. The data mining models themselves provide data leadership.
The term data mining I have discussed on this blog and on this webpage. The word encompasses not just the machine learning algorithms, but also the entire data preparation process before posting a result for actionable decisions. Mathematically, much of the difference in results comes from accurate and refined data preparation (often called ETL) prior to data mining model creation.
The technology is SQL Server 2008 R2. The standard and higher versions of SQL Server since 2005 have had the data mining technology included as part of the license. The data mining is part of Analysis Services, and again, SQL Server Data Mining is NOT an application but is instead a service. Because it is a service, it therefore needs a face, and that face could include SQL Server Management Studio (SSMS, included with SQL Server), Business Intelligence Development Studio (BIDS, also included with SQL Server) or as in this presentation, PowerShell (a free add-on from Microsoft often included with the newer operating systems, but also downloadable for many older Windows versions). Another face not discussed in this presentation is Excel 2007 or 2010 (both of which access data mining through the free data mining add-in). People can extend the product by making their own interface, and I discuss that topic in the presentation.
The demo code includes PowerShell, a BIDS project, and a SSMS code snipped (with DMX).
You can download the code from http://marktab.net/Resources/MicrosoftPresentations.aspx
Here is a picture showing the specific location:
Link to Microsoft’s Channel 9 website (which has my audio with the visuals of the slide deck and presentation: