Databases Reference
In-Depth Information
see results other than expected. This is mostly due to the way in which calcu-
lations are being applied. We highly recommend you know your cube design
well and the calculation precedence (order of calculations) and to subse-
quently verify the results. Several calculations can be applied to a measure:
semi-additive calculation, unary operator, and custom rollup. You learn more
about semi-additive measures briefly in this chapter and in detail in
Chapter
9
. When Analysis Services is evaluating the measure value for a member, it
initially calculates the regular aggregate of the measure value. This aggregate
can be Sum, Count, or any of the semi-additive functions (to be discussed
later). If a unary operator is specified, the unary operator rollup is applied for
the member across that specific dimension and the value of the measure is
overwritten. Finally, if a custom rollup is specified for the member, the value
resulting from the custom rollup MDX expression is evaluated as the final res-
ult. The evaluation of a cell value is done across each dimension and if di-
mensions have custom rollups and unary operators then all the unary operat-
ors are applied followed by custom rollups based on the order of the dimen-
sions within the cube.
Specifying Names to Levels of a Parent-Child Hierarchy
When you create multi-level hierarchies, various attribute hierarchies form the
levels of the hierarchy. For example, in a Geography hierarchy you will have
Country, State, City, and Zip Code as levels and when you browse the dimen-
sion you can see the names of the levels as the names of the attribute hier-
archies. Parent-child hierarchies are unique and different from the user hier-
archies. While creating regular hierarchies, you can see the various levels in
the dimension editor, but for parent-child hierarchies you cannot visually see
the number of levels unless you process and browse the hierarchy. The levels
within a parent-child hierarchy are embedded within the relationship. Analysis
Services allows you to define a name for each level of the parent-child hier-
archy. Typically, parent-child hierarchies can contain multiple levels and the
total number of levels corresponds to the depth of the parent-child hierarch-
ies. By default, Analysis Services 2005 provides names for the levels — Level
01, Level 02, and so on. Level N is based on the depth of the parent-child
hierarchy.
If you want custom names to be specified for each level, Analysis Services
2005 provides a property for just that. For example, if you have an org-struc-
Search WWH ::
Custom Search