MDX Essentials: Basic Set Functions: The Union() Function - Page 4

November 10, 2003

The behavior of the Union() function is demonstrated in its elimination of the duplicate Bellingham member - a characteristic that is the default of the primary syntax with which we have been working. We note, too, that the duplicate is deleted from the "tail"; the last of the duplicate pair is the one that is "dropped."

We can prove this to ourselves by overriding the default behavior, and using the ALL flag to allow the duplicate to be retained. To do so, we will make an adjustment to the query we created in Step 4 above.

{[Store].[All Stores].[USA].[WA].[Bellingham]})  

{[Store].[All Stores].[USA].[WA].[Bellingham]},ALL)

-- MDX13-5:  Tutorial Query Step 5

The Query pane appears as shown in Illustration 5.

The Results pane is populated, and the dataset shown in Illustration 6 appears. We see the "tail" instance of the duplicate Bellingham set is retained, thanks to the insertion of the ALL flag into our Union() function.

And so we see how to retain duplicates in a simple example that illustrates the influence of the ALL flag upon the Union() function. Let's turn now to the alternate syntaxes, both to examine them in general as options we might use, and to explore their implicit default behavior with regard to handling duplicates.

A blank query pane appears.

-- MDX13-6:  Tutorial Query Step 6

SELECT

{[Measures].[Warehouse Sales]} ON COLUMNS,

({[Store].[All Stores].[USA].[WA].Children} +

{[Store].[All Stores].[USA].[WA].[Bellingham]})

ON ROWS

FROM Warehouse

The Union() function is called into action this time using one of two alternate syntaxes. In this approach, we use the "+" operator to effect the Union() function, with a noteworthy characteristic of this arrangement being that the UNION keyword is not used at all. Let's take a look at the results we obtain through the use of this alternate approach.

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

Although the Books Online state that "Duplicated members are always retained when an alternate syntax is used," we see clearly in the foregoing results that the duplicate has been eliminated, just as it was using the primary syntax (default scenario.) Now let's attempt the same operation with the other alternative option in a final query.

A blank query pane appears.

-- MDX13-7:  Tutorial Query Step 7

SELECT

{[Measures].[Warehouse Sales]} ON
COLUMNS,

{{[Store].[All Stores].[USA].[WA].Children},

{[Store].[All Stores].[USA].[WA].[Bellingham]}}

ON ROWS

FROM Warehouse 

The Union() function is enacted in this query using the second of two alternate syntaxes. In this approach, we simply list the sets inside the braces, separated by a comma. Again, we note that the UNION keyword is not used at all. Let's see how duplicates are handled when performing a union via this alternate option.

The Results pane is populated, and the dataset shown in Illustration 8 appears.

In this case, we note that the duplicate, tail instance of Bellingham is retained, much as it was when using the ALL flag within the primary syntax example above.

In conclusion, we can see that the Union() function, both in its primary and alternate syntax forms, provides us the significant capability of performing combinations of sets. Our selection of an alternate syntax, or of the use of the optional ALL flag in the primary syntax, can also afford us flexibility in the handling of duplicates in queries we construct that use this function.

Summary ...

In this lesson, we explored the commonly used Union() function, whose purpose it is to combine two sets into one. We discussed the Union() function in general, and emphasized its provision for important capabilities within MDX, and its value within our analysis toolsets.

In addition to discussing the purpose and operation of the Union() function, we examined both primary and alternate syntaxes involved in its use, illustrating the application of each in practice exercises. Focusing on the treatment of duplicates in each of the syntactical approaches, we discussed the results we obtained in each exercise, remarking on distinguishing characteristics of each.

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

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