Data Mining Models

In this release, Microsoft® SQL Server™ 2000 Analysis Services introduces data mining, a new feature that integrates significant data analysis and prediction capabilities into Analysis Services. In addition to extensions to Analysis Manager, extensions to Decision Support Objects (DSO) and PivotTable® Service have also been made to support this powerful new feature.

• Analysis Services supports data mining from both relational and multidimensional data sources. The Analysis Services algorithms can train data mining models with data from any relational data source that supports OLE DB access, as well as from multidimensional cubes created with Analysis Services.
  
  
• Extensibility allows Analysis Services to be used with third-party tools such as mining model viewer components, providing flexibility and enhancement.
  
  
• OLE DB support has been enhanced with the OLE DB for Data Mining specification.
  
  
• Analysis Manager has been upgraded with the addition of new wizards and tools to help design, create, train, and browse data mining models. Dimensions and virtual cubes can also be created based on OLAP data mining models.
  
  
• DSO has been extended with the addition of the MiningModel object and other ancillary support objects.
  
  
• Data mining is integrated into PivotTable Service following the same pattern that was used in Analysis Services. In general, a data mining model is treated like a cube by PivotTable Service: You can create local data mining models, retrieve information from server cubes, and so on.