Input Selection Tab (Mining Accuracy Chart View)

Use the Input Selection tab of the Mining Accuracy Chart designer to specify the source of the data that is used to test the model and build the accuracy chart.

For more information: Testing and Validation (Data Mining)

Options

  • Synchronize Prediction Columns and Values
    Select to coordinate the predictable attributes in the grid so that, even if they have a different name, they are derived from the same predictable mining structure column during model training.

    Note   This option is selected by default. You should only clear this box for cases in which you know that two mining structure columns derive from the same underlying relational or multi-dimensional source, and that the columns contain the same states or have been discretized in the same way.

  • Select predictable mining model columns to show in the lift chart
    A grid that contains columns to control which models are included in the lift chart and how they are used in the lift chart.

    Value

    Description

    Show

    Select the box next to the name of each predictable column in the mining model that you want to display in the chart.

    If the chart is too complex to view easily, clear the box next to one or more columns to simplify the chart.

    Note

    You cannot create an accuracy chart unless at least one column is selected.

    Mining Model

    Lists the mining models that are contained in the mining structure.

    Predictable Column Name

    Select a predictable column that is contained in the mining models that are used to create the lift chart.

    Predict Value

    Select a value for the predictable column. If you leave this blank, the lift chart predicts how well the model performs for all states of the predictable column.

  • Select data set to be used for Accuracy Chart
    An option group that contains three options for specifying accuracy test data.

    Value

    Description

    Use mining model test cases

    Use the testing set that was created when you partitioned the mining structure, and apply the filter that is defined on the model. For information about model filters, see Filters for Mining Models (Analysis Services - Data Mining)

    Use mining structure test cases

    Use the testing set that was created when you partitioned the mining structure.

    Specify a different data set

    Specify a table from an existing data source view to use as a test data set.

Filtering Options

If you select the option Specify a different data set, you can define a data source view and create filters to apply to that data. When you create a filter, you are creating a WHERE clause in the query that returns the test data from the data source view.

Note   You cannot specify a filter on the mining model by using the Input Selection tab. To create a model filter, click the Mining Models tab and edit the model properties.

If you did not create a holdout set for testing when you created the mining structure, you can select this option and then specify the original data source view as a test set. By using this workaround, you can also set filters on the original data set.

  • Specify Column Mapping
    Opens the Specify Column Mapping dialog box, where you select the data source, specify case and nested tables, and map external data columns to the mining structure columns.

    For more information, see Specify Column Mapping Dialog Box (Mining Accuracy Chart).

  • Filter Expression
    Displays the filter condition that you built by using the filter editors.

  • Open Filter Editor
    Opens the Data Set Filter dialog box, which lets you select external tables, and set conditions on case table columns, and the Filter dialog box, which helps you build conditions that apply to individual columns in the selected table, or to columns in nested tables.

See Also

Tasks

Apply a Filter to a Mining Model

Reference

Mining Accuracy Chart Designer (Data Mining)

Concepts

Filters for Mining Models (Analysis Services - Data Mining)

Other Resources

Testing and Validation Tasks and How-tos (Data Mining)