Readme for SQL Server Plug-In Algorithms


Updated: 17 July 2006
Microsoft provides a complete framework that enables third-party data mining providers to integrate the following data mining objects into Analysis Services:
  • Data mining algorithms, known as plug-in algorithms.
  • Data mining viewers, known as plug-in viewers.
  • Data Mining Web Controls Library, a library that extends data mining viewers to the Web. For more information, see Data Mining Web Controls Library in SQL Server 2005 Books Online. To help you understand how to implement data mining algorithms and viewers into Analysis Services, this readme provides instructions for the following samples:
Data Mining Algorithm Plug-InThis sample allows data mining algorithm developers to concentrate on the algorithm instead of on the implementation issues. Data Mining Viewer Plug-InThis sample allows third-party algorithm providers to use their own viewers to display their new discovered patterns. The documents and source code associated with each of these samples are located in the C:\Program Files\Microsoft SQL Server\90\Samples\Analysis Services\DataMining\SQL Server Data Mining Plug-In Algorithms folder. Included in this folder are tutorials that describe how to implement and use each of these samples.

Note:
By default, the samples are not installed. To install samples, run Setup and use the Advanced option as described in the topic Running Setup to Install AdventureWorks Sample Databases and Samples in SQL Server 2005 Books Online. An updated version of AdventureWorks database is also available for download on the Web page.
An updated version of the SQL Server 2005 Database Engine samples is also available for download on the Web page. |



Note:
SQL Server 2005 Business Intelligence Development Studio is not supported on computers that run the Itanium-based platform; the 64-bit development environment is supported on x64-based platforms. SQL Server 2005 samples can be modified and run on computers that run either x86 or x64-based platforms when databases used by the samples are deployed on an Itanium-based platform.



Scenario

A provider wants to integrate a new data mining algorithm, which is named Pair-wise Linear Regression, into Analysis Services. After integrating this new algorithm into Analysis Services, the provider wants add a new viewer for the Pair-wise Linear Regression algorithm.

Languages

  • C++, for the algorithms.
  • C#, for the viewer.

Features

Algorithm Plug-In

Interface Explanation
IDMAlgorithm Implements both a model-producing learning algorithm and the prediction operations of the resulting model.
IDMAlgorithmNavigation Enables browsers to access the content of your model.
IDMPersist Enables the models learned by your algorithm to be saved and loaded by Analysis Services.
IDMAlgorithmMetadata Describes the capabilities and input parameters of your learning algorithm.
IDMAlgorithmFactory Creates instances of the objects that implement the algorithm interface and provides Analysis Services access to the algorithm metadata interface.


Viewer Plug-In

Interface Explanation
IMiningModelViewerControl Implements the viewer.


Prerequisites

Before you run this sample, you should have the following software is installed:
  • Microsoft SQL Server 2005, which includes the following components:
  • SQL Server 2005 Database Engine
  • Analysis Services
  • SQL Server Management Studio
  • Business Intelligence Development Studio
  • Two sample databases—AdventureWorksDW (data warehouse) and AdventureWorks (OLTP)
  • Analysis Services Samples
  • Adventure Works DW sample Analysis Services project, deployed to an instance of Analysis Services
  • An updated version of AdventureWorks database is also available for download on the SQL Server Developer Center Web page.
  • An updated version of the SQL Server 2005 Database Engine samples is also available for download on the SQL Server Developer Center Web page.

Building the Samples

Before building the plug-in samples, you must deploy both the AdventureWorksDW (data warehouse) and AdventureWorks (OLTP) databases.
To deploy the Adventure Works DW and AdventureWorks sample databases
  1. In Windows Explorer, open C:\Program Files\Microsoft SQL Server\90\Tools\Samples, and then double-click the SQLServerSamples.msi file.
  2. In Windows Explorer, open C:\Program Files\Microsoft SQL Server\90\Tools\Samples\AdventureWorks Analysis Services Project, open the folder corresponding to the version of SQL Server that has been installed, and then double-click the Adventure Works.sln file. The Adventure Works DW project opens in Business Intelligence Development Studio.
  3. Press F5 to build and deploy the project. Note: Building and deploying the AdventureWorksAS project deploys the mining model to the server.After you deploy the databases, you can implement the algorithm and viewer plug-ins.
To implement and use the data mining algorithm plug-ins
  1. In Windows Explorer, open C:\Program Files\Microsoft SQL Server\90\Samples\Analysis Services\DataMining\SQL Server Data Mining Plug-In Algorithms, and then double-click the SQL Server Data Mining Plug-In Algorithms Tutorial.doc file. Note: Implementing the plug-in sample might be easier for you if you print the tutorial and then use the printout to complete the tutorial.
  2. To build the Pair-wise Linear Regression Algorithm plug-in, complete Chapter 6, "Building a Shell Plug-In Algorithm" in the tutorial.
  3. To customize the Pair-wise Linear Regression Algorithm plug-in, complete Chapter 7, "Customizing the Algorithm: Pair-wise Linear Regression" in the tutorial.
  4. To use the customized Pair-wise Linear Regression Algorithm plug-in, complete Chapter 8, "Using the Customized Plug-In Algorithm" in the tutorial.
To implement the proprietary viewer
  1. In Windows Explorer, locate C:\Program Files\Microsoft SQL Server\90\Samples\Analysis Services\DataMining\SQL Server Data Mining Plug-In Algorithms, and then double-click the SQL Server Data Mining Plug-In Viewers Tutorial.doc file. Note: Implementing the plug-in sample might be easier for you if you print the tutorial and then use the printout to complete the tutorial.
  2. Complete all steps in the tutorial.

Comments

To obtain a better understanding about how to implement plug-in algorithms, read the file, SQL Server Data Mining Plug-In Algorithms Whitepaper.doc, which is included with the sample tutorials.

Change History


Release History
17 July 2006 New content: Added note on availability for Itanium-based platform.


© 2007 Microsoft Corporation. All rights reserved.

Last edited Nov 6, 2008 at 11:17 PM by bonniefe, version 16

Comments

No comments yet.