Step 1 - Define Dimensions

Dimensions are independent variables along which you wish to view information. Candidate dimensions are fields in a relational database or flat file table. This screen shows the continuous numeric variable "Capital" quantized into 20 intervals. Highlander automatically determined the intervals based on a best-fit algorithm. The designer can override this algorithm and specify the intervals. In addition, he can specify the formatting and labeling. The Highlander Explorer can apply a log transformation to compress extreme variation prior to quantizing the data. Numeric dimensions can also be self-calibrating, meaning that when the database is compiled, the quantizing intervals are automatically determined by scanning all tables and selecting a best fit interval set. This insulates the data mart somewhat from volatile source data.

In addition to continuous numeric data, other data types are supported for dimensions, including categorical lookup fields, temporal measures, uniform numeric ranges, and virtual data calculated from the other fields. Categorical dimensions can read an entity definition table at compile time so the dimension universe can be set on-the-fly based on current data. Temporal dimensions can gather at days, weeks, months, quarters, or years.