Specify the Training Data (Data Mining Wizard)
Use the Specify the Training Data page to identify how to use the columns when training the model.
Note
Columns can be both predictable and input at the same time.
For More Information:Mining Structures (Analysis Services), Mining Model Columns, Data Mining Wizard, Creating a New Mining Structure
Options
- Tables/Columns
Displays the tables and columns that you selected on the previous page of the wizard.
Key
Select to use the column as a unique identifier for the data.A column in a nested table that is marked as Key indicates the identifier of a row in the context of the associated case. If you also mark a nested table as predictable, the entire nested table becomes predictable. If no columns in the nested table are marked as input or predictable, the nested table will appear in the mining structure, but will be ignored in the model.
- Input
Select to use the column in creating predictions.
- Predictable
Select to enable the table or column to be predicted based on additional future input.
- Suggest
Click to open the Suggest Related Columns dialog box, which performs an analysis on a sample of data to identify input columns that show the greatest relationship to the selected Predictable column based on entropy. This analysis also applies to nested table columns or mining structures that are based on OLAP sources.
See Also
Reference
Data Mining Wizard F1 Help (SSAS)
Suggest Related Columns (Data Mining Wizard)
Specify Table Types (Data Mining Wizard)
Specify the Column's Content and Data Type (Data Mining Wizard)