Guidelines for Creating a Dataset Report

Before creating the dataset report, you need to make sure the attributes and metrics to be used as predictive inputs in the dataset report can also be used as inputs to the predictive metric. Recall that the predictive metric is the metric created from the predictive model after it is imported into MicroStrategy. By defining predictive inputs properly when you build the dataset report, you are guaranteed that the predictive model developed from that data receives the same inputs regardless of how that model is used in MicroStrategy.

You can also use an MDX cube as the dataset for your predictive model. This allows you to perform predictive analysis for your MDX cube data. The same guidelines for dataset reports listed below also apply to using MDX cubes as a dataset for your predictive model. For steps to integrate MDX cubes into MicroStrategy, see the MDX Cube Reporting Help.

The following guidelines are intended to help you create your dataset report.

• Use a flat report template.

  In a flat report template, attributes must be placed on rows and metrics on columns. Since the dataset report is a table consisting of a single header row followed by rows of data, placing attributes in the columns usually creates multiple header rows on each column, which cannot be easily represented in a database table.

• Use attributes and metrics as predictive inputs.

  To create predictive metrics, you can use both metrics and attributes as inputs, as described in:

  • Inputs for Predictive Metrics
  • Inputs for Predictive Metrics
  • Inputs for Predictive Metrics

    You can use metrics and attributes as inputs for predictive metrics by creating a training metric (see Creating a Predictive Model Using MicroStrategy).

• Match each metric's level with the attribute used on the rows of the dataset report.

  The attributes on the rows of the dataset report control the level, or dimensionality, of the dataset report. If a metric is used in the predictive model without a level, the metric results change based on the attributes of the report using the predictive metric. Creating another type metric for the predictive metric, which sets the metric level, can resolve this problem. For more information, see Inputs for Predictive Metrics.

• Set a condition on the metric that provides the proper grouping, when grouping a metric's results by an attribute.

  A metric condition is essentially a filter that allows an attribute to qualify metrics. For example, you can display customer revenue by payment method. For more information, see Inputs for Predictive Metrics.

• Set the attribute join type for the report.

  Setting the attribute join type ensures that any missing data does not cause rows to be deleted. To do this, complete the following procedure.

To Set the Attribute Join Type for the Report

1. Open the report in the Report Editor.
2. Choose Data > Report Data Options.
3. Under Categories, expand Calculations, and select Attribute Join Type.
4. Clear the Use Defaults check box.
5. Set the attribute join type of the report to either of the following:
• Preserve lookup table elements joined to the final pass result table based on the template attributes with filter keeps all attribute elements and applies all related filtering conditions.
• Preserve lookup table elements joined to the final pass result table based on the template attributes without filter keeps all attribute elements and ignores all related filtering conditions.
6. Click OK.