A dimension is a way of describing data within cubes. They represent the ‘who’, ‘what’, ‘where’ and ‘when’ of the data. A dimension may also contain the ‘how many, ‘how much’, ‘how long’ and so on.
Examples of dimensions include Cost Centres, Accounts, Products, Stores, Cities, Customers, Suppliers and so on.
Dimensions contain elements. Elements can be a simple list or may add up to a total or a complicated hierarchy with many sub-totals (called roll-ups). Elements can be included in several sub-totals, therefore data can be added up in lots of different ways.
By bringing two or more dimensions together, you can create a data cube. Inside the cube the intersection of each single element in each dimension can hold a value. Therefore, every piece of data in the cube has its own set of unique co-ordinates relating to the elements in each dimension in the cube. We can manipulate our view of the dimensions in the cube to present information in different ways.
The data will naturally total-up through the dimension hierarchies. This allows us to look at summarised data, but easily drill-down through the dimensional hierarchies to reveal more detail.
There are a few different types of dimensions which are worth noting, although IBM Planning Analytics treats them all the same:
Contains the elements to record, group and analyse data over a time-period, which are often days, weeks, months, quarters, years.
Version or Scenario Dimension
Comprises elements for data type such as Actual, Budget, Forecast for example, and often the variances between the data types.
Contains the values that are being measured, such as quantity, price, unit cost, value or count.