Further Combination of BottomCount() with Other MDX Functions - Page 2

November 28, 2008

The calculated member we create with this MDX is a “busy place.” It accomplishes the BottomCount(), providing as the << Set >> argument a Descendants() function. We saw the BottomCount() / Descendants combination in very similar action in the practice session we undertook within Combine BottomCount() with Other MDX Functions to Add Sophistication, and so I won’t go into that any further here, other than to point out that the function is structured to return the set of the members of the Geography dimension (Geography hierarchy) that belong to the level it specifies, the State-Province level. It is from this level that we are able to obtain the string representing the U.S. State Name that we need.

The BottomCount() function is provided a << Count >> argument of “1” , obviously, as we seek to return the “worst” (or “least”) performing State with regard to the Reseller Order Quantity measure, the final argument of the function, << Numeric Expression >>. The interim result of the BottomCount() function, therefore, is to return a single “worst” performing State member for each Month, which enables us to leverage the .Item function. .Item, in turn allows us to exploit its capability to return the member in a tuple, based upon its position. By using .Item(0), we select the first single-member (the product of the BottomCount() of 1) tuple in each Month set. (.Item contains a zero-based specification of position, hence the zero as the position argument.) We then use .Item(0) again, immediately after the first .Item function, to select the first (and only) member in the single-member tuple.

Finally, we use the .Name function to return the name of the State member that we have so precisely isolated. This function, combined with the functions just described, result in a calculated member that returns the name of the State with the bottom / lowest shipping Reseller Order Quantity in a given Month in CY 2003, which is handily identified by the BottomCount() / Descendants() combination we discussed earlier.

 {[Measures].[Reseller Order Quantity], 
    [Measures].[Least Reseller Order Qty State]} ON AXIS(0),

The Query pane appears as depicted in Illustration 3, with our modifications marked.

The Results pane is populated, and the resulting dataset, partially shown in Illustration 4, appears.

We note that the Name of a single U.S. State appears for each Month, alongside the respective Reseller Order Quantity totals. The presence of the months of CY 2003, which populate the rows already, result in the evaluation of the “Least Reseller Order Qty State” calculated member for each month respectively, giving us an outcome that meets the expressed requirements of the information consumers.

Using the results from our latest query execution, we can easily verify that the calculated member is producing the results that its name purports. An example of such verification appears in Illustration 5, where we see that, for November 2003, the “least performing” U. S. State returned by our calculated member reflects the name of New Mexico that displays the “bottom” total value in the monthly Reseller Order Quantity measure at its left.

(November 2003 as Example Month)

In the example, we can easily see that New Mexico, which displays a “(null)” total Reseller Order Quantity for the Calendar Year period chosen as an example, November 2003, is, indeed, the “bottom” performer. First, “no” (that is, “null”) Reseller Order Quantity is clearly at the bottom of the list, which contains members with “non-zero” values. And among other “nulls” in the list, we see that New Mexico, as a member, lies at the “bottom” of the “natural order” of the group of other States returning nulls – the default behavior of the BottomCount() function when “ties” occur in the “least” performer.

All that remains, to align the results dataset with the precise presentation requirements specified by the information consumers, is to remove the Reseller Order Quantity measure. (Recall that we added this measure earlier simply to allow us to verify the correct operation of the “Least Reseller Order Qty State” calculated member.)

-- MDX073-3, Alignment of Results Dataset to the Business Requirements

 {[Measures].[Least Reseller Order Qty State]} ON AXIS(0),

 NON EMPTY

 CROSSJOIN
 (

 {
         DESCENDANTS( [Geography].[Geography].[United States],
            [Geography].[Geography].[State-Province] )
         }

To summarize the final set of modifications, the query appears, after the addition of the most recent comment (Step 14 above), and before our other modifications, as follows, with syntax to be deleted enclosed in rectangles:

 -- MDX073-3, Alignment of Results Dataset to the Business Requirements
 
 WITH 
 MEMBER
      
    [Measures].[Least Reseller Order Qty State]
 
 AS
    'BOTTOMCOUNT
       ( 
          DESCENDANTS
             ( 
                [Geography].[Geography].[United States],
                   [Geography].[Geography].[State-Province] ), 
                      1, ([Date].[Calendar].CURRENTMEMBER,   
                   [Measures].[Reseller Order Quantity]
                   ) 
                ).ITEM(0).ITEM(0).NAME'
 
 SELECT 
   {
[Measures].[Reseller Order Quantity],
 
     [Measures].[Least Reseller Order Qty State]} 
   ON AXIS(0),
 
   
 
NON EMPTY
   CROSSJOIN
      (
         {
            DESCENDANTS([Date].[Calendar].[Calendar Year].&[2003], 
               [Date].[Calendar].[Month])
            
},
             
         
{
            DESCENDANTS( [Geography].[Geography].[United States],
               [Geography].[Geography].[State-Province] )
            }
         )
      ON AXIS(1)
 
 FROM 
    [Adventure Works]

The Query pane appears, with modifications having been made, as depicted in Illustration 6.

The Results pane is populated, and the dataset shown in Illustration 7 appears.

Summary ...

This article served as a conclusion of our multi-part examination of the BottomCount() function, which we began, and continued, with the two previous articles of the MDX Essentials series, Basic Set Functions: The BottomCount() Function and Combine BottomCount() with Other MDX Functions to Add Sophistication, respectively. We briefly reviewed the operation of the BottomCount() function, and then extended our examination of BottomCount(), yet again, to include another somewhat more sophisticated use.

As we routinely do within the articles of the MDX Essentials series, we defined an illustrative business need, as posed to us by hypothetical groups of information consumers. We next discussed the need in general, with regard to challenges inherent to meeting them with the BottomCount() function in particular. We then constructed, in a step-by-step manner, a final query to meet the expressed requirements using a combination of BottomCount() and other MDX functions we have explored within articles of the MDX Essentials series. Throughout our practice example, we discussed the syntax contained within the solutions we constructed for the information consumers, as well as the results datasets we obtained in employing a combination of BottomCount() and other functions, together with other surrounding considerations.

About the MDX Essentials Series ...

This article is a member of the MDX Essentials series, a monthly column designed to provide hands-on application of the fundamentals of the Multidimensional Expressions (MDX) language, with each article progressively adding features designed to meet specific real-world needs.

For more information about the column in general, as well as the software and systems requirements for getting the most out of the lessons included, please see my first article, MDX at First Glance: Introduction to MDX Essentials, among others.

» See All Articles by Columnist William E. Pearson, III

Discuss this article in the MSSQL Server 2000 Analysis Services and MDX Topics Forum.