Data Mining Designer
Data Mining Designer is the primary environment in which you work with mining models in Microsoft SQL Server 2005 Analysis Services (SSAS). You can access the designer either by selecting an existing mining structure item or by using the Data Mining Wizard to create a new mining structure and mining model. You can use Data Mining Designer to perform the following tasks:
- Modify the mining structure and the mining model that were initially created by the Data Mining Wizard.
- Create new models based on an existing mining structure.
- Train and browse mining models.
- Compare models by using accuracy charts.
- Create prediction queries based on mining models.
A mining structure item contains a single mining structure and all its associated mining models. Each mining model can have a different algorithm type, parameter setting, and columns that are included from the mining structure. Because associated models are all contained within a single structure, you can compare the performance of the models by using viewers and accuracy charts.
The following sections describe the individual tabs in Data Mining Designer.
Mining Structure Tab
Every time that you add a new mining model to your project, you use the Data Mining Wizard to define a new mining structure that contains the columns that the associated mining models use. Data Mining Designer opens to display the Mining Structure tab, which has two parts: the data source view and the columns in the structure. The Data Source View pane displays the tables that the data source view contains. You can also use this pane to explore the underlying data that is contained within the view, and to access Data Source Designer. The left pane lists, in a tree view, the columns in the selected mining structure. You can modify the mining structure by adding or removing columns and nested tables. You can modify column properties by selecting a column and using the Properties window.
You can also use this tab to process the mining structure and its associated mining models.
Mining Models Tab
Use the Mining Models tab to manage existing mining models and to create new models. Mining models are based on a mining structure that is displayed in the Mining Structure tab and which you defined by using the Data Mining Wizard. The main surface area of the Mining Models tab contains a grid with rows for each column in the mining structure. The first column in the grid displays the names of the columns in the mining structure. Each additional column in the grid represents a mining model that is associated with the structure. Each row in a mining model column describes how the model uses the mining structure column that is associated with the row.
In the Mining Models tab, you can change the algorithm type, add or remove columns that are associated with the model structure, adjust algorithm-specific column properties, specify the mining model column usage, and adjust algorithm parameters that are associated with the mining model. You can also process the mining structure together with selected models or with all the associated models.
Mining Model Viewer Tab
Use the Mining Model Viewer tab to visually explore your mining models. Each mining model is associated with a custom viewer that displays content that is specific to that model. You can also view mining model content by using the content viewer.
Mining Accuracy Chart Tab
Use the Mining Accuracy Chart tab to test the predictive accuracy of a single mining model, or to compare the effectiveness of multiple mining models contained within a mining structure. The tab contains tools for filtering the data, selecting mining models, and displaying the results in a lift chart, profit chart, or classification matrix.
Mining Model Prediction Tab
The Mining Model Prediction tab includes Prediction Query Builder, which you can use to create a Data Mining Extensions (DMX) prediction query. The tab contains tools for specifying mining models and input tables, mapping the columns in the mining model to columns in the input table, adding functions to a query, and specifying criteria for each column.
After you design a query, you can use different views in the tab to display the results of the query and to modify the query manually. You can also save the results of the query to a table in a database.