When you use Data Mining Extensions (DMX) to query objects in Microsoft SQL Server Analysis Services, you can use functions to return more information than just the values in the columns in the data mining model or input dataset. For example, you can use DMX queries to return not only the prediction value of a column, but also the probability that the prediction is correct. You can use not only DMX functions, but also functions from Microsoft Visual Basic for Applications (VBA), Microsoft Excel, and stored procedures.
You can use DMX functions to perform the following tasks:
Return statistics about a prediction such as the probability and support.
Filter your query results.
Reorder a table expression.
Most DMX functions return a scalar value, such as the support for a prediction, but some return a tabular result. For example, the PredictHistogram function returns a table that contains the support and probability for each state of the specified predictable column. The results are displayed as a new tabular column.
In addition to DMX functions, you can also call a variety of VBA and Excel functions from DMX statements. For example, you can use the lCase function to modify how the Attribute_Name column in the TM_Decision_Tree model content is displayed. This is shown in the following code sample.
SELECT lCase([Attribute_Name]) FROM [TM_Decision_Tree].CONTENT
If the same function exists in both VBA and Excel, you must prefix the function name in your DMX statement with either VBA or Excel. For example, you would use VBA!Log or Excel!Log. If the VBA or Excel function that you want to use also exists in DMX or Multidimensional Expressions (MDX), or if the function contains a dollar sign character ($), you must use square brackets () to escape the function. For example, the function call would be [VBA!Format].
You can use common language runtime programming languages to create stored procedures that extend the functionality of DMX. For example, a regression tree mining model returns coefficients, such as A, B, and so on, that describe the regression equation, but the model does not return the equation itself, such as A+Bx = y. However, you can write a stored procedure that uses the data mining model object to navigate the content schema, and to return the regression equation as an output. Therefore, a DMX statement can return a list of the regression equations as part of a query result.
For More Information: Multidimensional Model Assemblies Management