Level
A level is a step of aggregation within a hierarchy. You can think of a level as being analogous to a rung of a ladder. Just as you can move up and down a ladder by using its rungs, you can move up and down into the hierarchy of data by using its levels. The first level represents the highest step in the hierarchy; the second level is the child of this first level, and the third level is the child of the second level, and so on. Figure 1 illustrates a hierarchy in which there are 3 levels shown, namely the Location, Continent, and Country levels.
Member
A member is an element of a dimension. The member element may represent one, or more, occurrences of data within a dimension. A member is the lowest form of reference used to access cell data. It is useful to think of a member as a child entity that belongs to its parent dimension. For example, a member of the Location dimension might include North America, Europe, France, or the USA. In Figure 1, North America and Europe are members of the Location dimension. You can drill down into North America to view its members Canada, and USA. In this example, the members Canada, and USA are children of the North America member.
Measure
A measure is a value that describes a standard of measurement used for evaluating data. It is a specialized form of a dimension in which no hierarchy needs to be defined, and no levels are used. Examples of a measure include: Sales Count, Profit, or Number of Employees. Measures have meaning only when used in conjunction with one or more dimensions. Let's assume we have a measure (say, for instance "Amount of Sales") which has a value of $300. This information is meaningless to us because it lacks the context needed to add that meaning. However, if we define our measure with two other dimensions, for example, Time and Store ID, we can now interpret the data as:
Amount of Sales = $300 for the month of September at Store #232.
With the addition of these dimensions, this information has a rich element of meaning for us.
Attribute
An attribute can be thought of as a characteristic of a dimension with no inherent hierarchical relationships. Attributes add flexibility to multidimensional data because they can be organized into different hierarchies.
For example, suppose we have a Customer dimension with an attribute called "City". We can use this attribute in more than one Customer-related hierarchy (Country-State-City and/or City-Gender-Education). This gives us far more flexibility when building and analyzing multiple hierarchies.
Named Set
A named set represents a defined subset of dimensions and measures. These entities are most useful when trying to show subsets of data for multiple groups of users.
Multidimensional Data Cube Example
In this example, we have three major dimensions of data: Location, Time, and Product.
Note
We use a 3-D shape as a simplistic, visual representation of the multidimensional data. If the multidimensional data contains more than three dimensions, then the model must be extended beyond this 3-D representation.
Each data dimension is sub-divided into levels. For example, the Location dimension is divided into 3 levels. The first is the "continent" level, which contains two members: North America, and Europe. Then within each of these member levels, there are one or more child members at the "nation" level. In addition to this, there are two measures: Total Sales and Expenses. These measures exist at the intersection of the three data dimensions.
As displayed in Figure 2, some cells in the cube may not contain valid information.