Extending SQL Server Data Mining
Download
Report
Transcript Extending SQL Server Data Mining
Extending SQL Server Data
Mining
Ryan Ebert
Richard Rogers
Nick Crampton
Plugins
Use store procedures to add functions
past basic data mining
Use visual studio to enhance data mining
Develop plug in algorithms
Develop new viewers to visualize data
Plug-in algorithms
SQL server 2005 allows third parties to
develop their own algorithms for Analysis
services
You can add or disable algorithms
Add on algorithms appear to end user the
same as built in algorithms
How plug-in algorithm works
Receives stream of data and extracts and
stores patterns
Receives requests to describe patterns
and apply them to new data and make
predictions
Managed Plug-ins
Plug-in is a COM object
Possible to use COM interfaces with .net
Usable with languages such as C# and VB
Memory is not managed by Analysis
services
Installing Plug-in Algorithms
Register COM
Update server’s INI file
Instructions are listed in sample plug-in
Use Data Mining Viewers
Add viewers to algorithms
Add viewers to display your own
algorithms
Included controls are .NET Winform
controls
You can create your own viewing library
Conclusion
SQL server 2005 is very extendable
Depends on your ability to program
Works with plug-ins
Create models, predictions, content
navigation, data mining viewers