Managing Mining Structures in Data Mining Designer
The Mining Structure tab of Data Mining Designer displays a mining structure that you select in Solution Explorer. You can use this tab to modify mining structures that you create with the Data Mining Wizard. For more information about mining structures and other data mining objects, see Mining Structures (Analysis Services - Data Mining).
The Mining Structure tab is broken into the following panes:
These panes function differently depending on whether you are working with structures that are based on relational data sources or structures that are based on online analytical processing (OLAP) data sources.
Tree View Pane
You can interact with both relational and OLAP mining structures in the same way in the Tree View pane of Data Mining Designer, except that for OLAP structures you can also slice the source cube.
The tree view contains all the columns and nested tables that exist in the mining structure. The following table lists the tasks that you can perform in the Tree View pane, and directs you to topics that explain how to perform these tasks.
Add a mining structure column
Add a mining structure nested table
Process all models in a mining structure
Add a new model to a mining structure
When you add a new column to a mining structure, the Select a Column dialog box opens. After you add a column, you can use that column in any mining model in the structure. Models cannot use columns that have not been added to the structure. You can modify the properties of each column in a model by using the Properties window.
For More Information: Setting Properties on a Mining Model
When you add a new nested table to a mining structure, the Select a Nested Table Key Column dialog opens. This lets you select the key column that relates the nested table to the parent table in the mining structure.
When you add a new model to the structure, the New Mining Model dialog opens. To modify the new mining model after you create it, use the Mining Models tab of the designer.
For More Information: Managing Mining Models in Data Mining Designer
Slicing a Cube
You can filter the OLAP cube on which a mining structure is based by right-clicking the tree view and selecting Define Mining Structure Cube Slice. This opens the Slice Cube dialog box, which displays a grid of four columns: Dimension, Hierarchy, Operator, and Filter Expression.
You can set as many filters on the cube as are required to return the data that you need for the mining model. You can also define slices on individual cube slices. For example, if your structure contains two nested tables that are based on products, you could slice one table on March 2004 and the table other on April 2004. The resulting model could then be used to predict purchases made in April based on the purchases that were made in March.
Data Source View Pane
The Data Source View pane of Data Mining Designer contains different information depending on whether you are looking at a mining structure that is based on a relational data source or a mining structure that is based on an OLAP data source.
Relational Mining Structure
For mining structures that are based on a relational data source, the Data Source View pane contains basic functionality that you can use to explore the data and the relationships that exist in the data source view from which the mining structure is built. The following table lists the tasks that you can perform in the Data Source View pane, and directs you to topics that explain how to perform these tasks.
Access Data Source View Designer to modify the data source view
Right-clicking in any empty area within the pane displays a menu with several options for viewing the tables that are in the data source view. Selecting Show Tables opens the Show Tables dialog box, which you can use to filter the tables that are shown in the pane. You can also select Show All Tables to show all the tables in the data source view, or Show Only Used Tables to show only those tables that the mining structure uses. If the data source view contains many tables, you can use Find Table to locate a particular table. You can use Arrange Tables to modify the arrangement of the tables in the data source view. If you prefer to view the data source view in a tree format, point to Show Data Source View in and then select Tree.
Selecting Edit Data Source View opens Data Source View Designer, where you can perform such tasks as adding or removing tables, modifying relationships between tables, and creating named queries.
For More Information: Designing Data Source Views (Analysis Services)
Right-clicking a table in the Data Source View pane and selecting Explore Data opens the Explore <table> Table window. Within this window, you can use various charting options and table views to explore how the data is distributed within the table.
OLAP Mining Structure
For mining structures that are based on an OLAP data source, the Data Source View pane displays a view of the cube attributes and measures that exist in the source cube. If your structure contains a nested table, this view also displays the cube attributes and measures that are available for use by the cube. You can add columns to the mining structure by dragging them from the tables in the Data Source View pane into the appropriate location in the Tree View pane.