MicroStrategy ONE

Select Metrics

You can select the metrics and attributes from your project or MDX cube to be used in the training metric. To locate the metrics and attributes to be used, you can use the following:

  • Object Browser: This option is available for all types of training metrics. Use the Object Browser to browse your project for metrics and attributes to be used in the training metric. If you are creating MDX cube training metrics, you can only select metrics and attributes within the MDX cube.

  • Report Objects: This option is available if you are creating derived training metrics only. Use the Report Objects to select attributes and metrics from the report to be used in the training metric.

Once selected, metrics and attributes can be added to the following areas using a drag-and-drop method or by selecting the metric or attribute and clicking the arrow buttons.

If an attribute is selected as a input, the Training Metric Wizard automatically creates a metric expression to be used as input to the training metric. By default, this expression uses the following format:

Max(AttributeName){~}

The ID form of the attribute is used as an input to the training metric. If a different attribute form should be used as an input, a metric must be manually created prior to creating the training metric. For steps on how to create this metric, see Creating metrics to use additional attribute forms as inputs for predictive metrics.

The options that are available are dependent on the type of predictive analysis that is being used for the training metric:

  • Dependent Metric: Add the metric or attribute representing the field value to be predicted by the model. There must be one and only one dependent metric or attribute. This option does not apply to cluster analysis or association rules analysis.

  • Transaction Metric: This field appears only for association rules analysis, replacing the Dependent Metric field, and specifies how items will be grouped. There must be exactly one transaction metric or attribute.

  • Independent Metrics: Add metrics and attributes for each of the independent variables to be considered while generating the predictive model. There must be at least one independent metric or attribute. This option does not apply to association rules analysis.

  • Item Metric: This field appears only for association rules analysis, replacing the Independent Metrics field, and specifies the items to be associated. There must be exactly one item metric or attribute.

  • Segmentation Metrics: The Training Metric Wizard allows for the selection of Segmentation metrics or attributes for regression-type analysis. This option results in the generation of a mining model, which is used to support model selection. Add one metric or attribute to represent each level within the tree of the model. If there are no Segmentation metrics or attributes specified, a standard regression model will be generated. This option only applies to the three types of regression analysis.

  • Show advanced options: Select this check box to display the Advanced Options step of the Training Metric Wizard. This step allows you to define analysis-specific and model input-specific options.

Click Next to proceed with the Training Metric Wizard.