As we can see within the list that appears next, selection options for Discretization Method includes:
- None - the default;
- Automatic Analysis Services determines which discretization method to use;
- EqualAreas The algorithm divides the data into groups that contain an equal number of values. (This method is best used for normal distribution curves, but does not work well if the distribution includes a large number of values that occur in a narrow group in the continuous data);
- Clusters The algorithm divides the data into groups by sampling the training data, initializing to a number of random points, and then running several iterations of the Microsoft Clustering algorithm using the Expectation Maximization (EM) clustering method. (The Clusters method can only be used with numeric columns.) The Clusters method is useful because it works on any distribution curve, although it requires more processing time than the other discretization methods.
Advanced Property: EstimatedCount
We use EstimatedCount to specify the estimated number of members within the attribute. This value defaults to zero, until we first run the Aggregation Design Wizard. Thereafter, the number is either the amount last counted by Analysis Services, or a user-provided estimate of the member count. That is, we can either manually enter an estimated value, or we can allow the wizard to count the number of records for us. (We can enter a value manually in those cases where we know the number, and want to save the time it takes to query the database for the count.)
Advanced Property: IsAggregatable
IsAggregatable allows us to specify whether the values of the attribute members can be aggregated within a hierarchy. The default value is True, which means that the attribute hierarchy contains an (All) level. If the value for this property is False, the attribute hierarchy does not contain an (All) level.
Advanced Property: OrderBy
The OrderBy property affords us a means of describing how the members contained in the attribute hierarchy should be ordered. The default value is Name, which specifies that ordering of the attribute members is based on the value of the NameColumn property, if any. If the NameColumn property is empty, members are ordered by the value of the key column.
7. Click the selector button (the downward pointing arrow shown circled in Illustration 11) to the right of the setting box for the OrderBy property.
Illustration 11: Click the Selector Button to the Right of the OrderBy Property
As we can see within the list that appears next, the entire set of options that we can specify as a basis for ordering includes the following:
- Key - Order by the value of the key column of the attribute member;
- Name - Order by the value of the NameColumn property (the default, as we have noted);
- AttributeKey - Order by the value of the member key of a specified attribute, which must have an attribute relationship to the attribute;
- AttributeName - Order by the value of the member name of a specified attribute, which must have an attribute relationship to the attribute.
Advanced Property: OrderByAttribute
The OrderByAttribute property allows us to identify the attribute by which we wish to order the members of the attribute hierarchy (assuming we have selected AttributeKey or AttributeName within the OrderBy property discussed above).
8. Click the selector button (the downward pointing arrow) to the right of the setting box for the OrderByAttribute property, shown circled in Illustration 12.
Illustration 12: Click the Selector Button to the Right of the OrderByAttribute Property
As we can see from the list of options that appears, we can order by attributes which have an attribute relationship with the attribute we are ordering (in this case, Geography Key).
We will extend our introductory examination of dimension attributes, specifically continuing our discussion with the Basic, Misc, Parent-Child and Source groups of properties, within Part II of this article.
NOTE: Please consider saving the project we have created to this point for use in subsequent related articles of this subseries, so as to avoid the need to repeat the preparation process we have undertaken initially, to provide a practice environment.
9. Select File -> Save All to save our work, up to this point, within the originally chosen location, where it can be easily accessed for our activities within subsequent articles of this subseries.
10. Select File -> Exit to leave the design environment, when ready, and to close the Business Intelligence Development Studio.
In this, the first half of a two-part article introducing dimension attributes, we continued our current subseries focusing upon dimensional model components, with an objective of discussing the associated concepts, and of providing hands-on exposure to the properties supporting each. We reviewed our initial introduction to the dimensional model and summarized its role in meeting the primary objectives of business intelligence. Next, we provided a brief overview of dimension attributes in general.
Having covered the general characteristics and purposes of attributes, we began our focus upon the properties underlying them, based upon the examination of a representative attribute within our sample cube. In this article, we discussed the Advanced group of properties. Finally, we looked forward to Part II of this article, where we explore the remaining attribute properties, those belonging to the Basic, Misc, Parent-Child and Source groups.
» See All Articles by Columnist William E. Pearson, III
Discuss this article in the MSSQL Server 2000 Analysis Services and MDX Topics Forum.