Cubes

IBM® Planning Analytics stores the data that you need for planning and analysis in cubes.

Each cube typically has a specific purpose. Suppose that you are analyzing sales; you have a cube that measures the sales for Sedan cars over time. The cube contains three dimensions: Measures, Product, and Month. Each measure, such as Sales, is organized by a product and a month. For example, the cell value 300000 represents the sales of Sedan-1 in the month of January (Jan).

Example of a cube with three dimensions

A cube has two or more dimensions. The number of dimensions that there are in a cube depends on the purpose of a cube. For example, a two-dimensional cube is useful as a lookup table; you can store exchange rates in a lookup table.

A revenue planning cube might have the following dimensions: Products, Versions, Regions, Measures, Time.

For a thorough review of cube design best practices, see A Best Practice Guide to Designing TM1 Cubes on the IBM Accelerator Catalog.

Guidelines for designing cubes

  1. List the measures you want to track in your business analysis. Examples of measures include sales amounts, units sold, expenses, acquisition values, and campaign costs.
  2. Decide what dimensions you need. Consider the following questions:
    • In most analyses, you track measures over time. What is the base time interval: days, weeks, months?
    • Is there a geographic dimension?
    • Do the measures vary by customer and product?
    • Is there a scenario, or versions dimension. For example, do you want to see the actual figures versus the budget?
  3. Determine the hierarchical structure of your dimensions. Every dimension has at least one hierarchy. If multiple hierarchies are enabled, you can use them to see alternative rollups of the data in the same view.
  4. Create a list of attributes you want to associate with the dimension members. Examples of attributes include store square footage, customer IDs, and local language versions of member names.
  5. Define the display formats for the measures in your cubes. For example, define Gross Margin as a percentage and Sales as a currency amount.
  6. The dimension order in a cube is important. The order that you select the dimensions when you create a cube can impact performance. In general, select the dimensions in order of the smallest, sparse dimension first, and the largest, dense dimension last. A dense dimension has a high percentage of values for its members. However, you might find that it is better to put a small dense dimension, such as a versions dimension, before a large, sparse dimension such as a products dimension.
  7. If the dimension contains members that have the text type, or is a pick list, this dimension must be last in the cube. Text type members and pick list members must be in the same dimension.