Dela via


Managing Data Mining Structures and Models

The management tasks that you can perform on existing mining structures and mining models depend on the client that you use to access the server. In Business Intelligence Development Studio, you can work with underlying data, rename or change objects, process objects, and deploy solutions. In SQL Server Management Studio, you can reprocess objects that have been deployed, browse models, build prediction queries, measure the accuracy of mining models by using charts and other tools, and delete existing objects; however, you cannot create new objects, rename objects, or change their properties. 

In addition to Business Intelligence Development Studio and SQL Server Management Studio, you can manage data mining objects programmatically, by using AMO, or use other clients that connect to an Analysis Services database, such as the Data Mining Add-ins for Microsoft Excel 2007.

This section explains how to manage existing mining structures and models by using Business Intelligence Development Studio and SQL Server Management Studio to make the following changes:

Mining structures

  • Changing the definition of the mining structure by adding columns or changing column properties.

  • Adding a new model to a structure

  • Deleting or reprocessing a mining structure.

  • Deleting cached data that is related to the mining structure.

  • Exporting the mining structure definition to a file.

Mining models

  • Changing the definition of the mining model by adding columns, changing column definitions, changing the usage of columns within the model.

  • Changing the algorithm that is associated with the model, or changing algorithm parameters.

  • Enabling drillthrough.

  • Adding a filter.

    Note   All changes to a mining structure or mining model, including changes to metadata such as a name or description, require that the structure or model be reprocessed.

Location of Mining Structures and Mining Models

Mining structures and models that have been processed are typically stored in an instance of Analysis Services.

If you create a connection to an Analysis Services database in Immediate mode when developing your data mining objects, any objects that you create are immediately added to the server as you work. However, if you design data mining objects in Offline mode, which is the default when you work in Business Intelligence Development Studio, the mining objects that you create are only metadata containers until you deploy them to an instance of Analysis Services. Therefore, any time that you make a change to an object, you must redeploy the object to the Analysis Services server. For more information about data mining architecture, see Physical Architecture (Analysis Services - Data Mining).

Note

Some clients, such as the Data Mining Add-ins for Microsoft Excel 2007, also let you create session mining models and mining structures, which use a connection to an instance but store the mining structure and models on the server only for the duration of the session. You can still manage these models through the client, the same as you would structures and models stored in an Analysis Services database, but the objects are not persisted after you disconnect from the instance of Analysis Services.

Managing Structures and Models in Business Intelligence Development Studio

Business Intelligence Development Studio offers many features that make it easy to browse, edit, and manage data mining objects. For an overview of how to work with projects in Business Intelligence Development Studio, see Introducing Business Intelligence Development Studio

Managing Structure and Models in SQL Server Management Studio

In SQL Server Management Studio, you can script, process, or delete mining structures and mining models. You can view only a limited set of properties by using Object Explorer; however, you can view additional metadata about mining models by opening a DMX Query Editor window and selecting a mining structure.

Managing Structures and Models Programmatically

You can create, alter, process, and delete data mining objects by using the following programming languages. Each language is designed for different tasks and as a result, there might be restrictions on the type of management operations that you can perform. For example, some properties of data mining objects cannot be changed by using Data Mining Extensions (DMX); you must use XML/A or AMO.

Analysis Management Objects (AMO)

An object model built on top of XML/A that gives you full control over data mining objects. Lets you create, deploy, and monitor mining structures and mining models

Restrictions: None.

Data Mining Extensions (DMX)

Can be used with other command interfaces such as ADO.NET or ADOMD.Net to create, delete, and query mining structures and mining models.

Restrictions: Some properties cannot be changed by using DMX.

XML for Analysis (XML/A)

Gives you full control over data mining objects. All management operations between the client and the server can be performed by using XMLA. For convenience, you can use the Analysis Services Scripting Language (ASSL) to wrap the XML.

Restrictions: Business Intelligence Development Studio generates some XML statements that are supported for internal use only, and cannot be used in XML DDL scripts.