Dimensions and Separation
When a cube is created from a data table, any column of the table can be designated as a dimension, in which case its records are organized into groups that contain like values. In the cube, this dimension exists as an array of distinct values (the values of the dimension), each of which points to the set of records associated with that value.
Definitions of "dimension" can become very theoretical in so-called "multi-dimensional analysis", but in Antaeus none of this theory is used. A dimension is a strictly functional concept, which can be visualized using a couple of interrelated tables.
For instance, here is a simple table of nine records containing one dimension, Color, and three measures, X, Y, and Z:
| Record | Color | X | Y | Z |
| 1 | Red | 9 | 2 | 5 |
| 2 | Green | 4 | 5 | 2 |
| 3 | Blue | 10 | 2 | 10 |
| 4 | Red | 3 | 4 | 3 |
| 5 | Green | 4 | 11 | 9 |
| 6 | Blue | 1 | 8 | 4 |
| 7 | Green | 5 | 8 | 12 |
| 8 | Blue | 11 | 4 | 3 |
| 9 | Red | 4 | 2 | 8 |
In Antaeus, a cube built from this table would contain a dimension array called Color, containing the three values "Red", "Green", and "Blue". These would each point to three records each, as exemplified in the following table:
| Color | Records |
| Red | 1, 4, 9 |
| Green | 2, 5, 7 |
| Blue | 3, 6, 8 |
This means that, say, "Green" points to the following subset of records:
| Record | Color | X | Y | Z |
| 2 | Green | 4 | 5 | 2 |
| 5 | Green | 4 | 11 | 9 |
| 7 | Green | 5 | 8 | 12 |
This is the extent to which Antaeus conceptualizes dimensions. And though this may seem basic, its applications are boundless. This is all that is needed to accomplish a great deal of work.
A cube can contain however many dimensions are needed, with however many values are necessary. The values of a dimension can be separated so that the records, shown in scatter plots as data points, can be easily distinguished from one another using color. We call this functionality separation.
To make it clear as possible, we'll illustrate this using the College Data demo cube, which is installed with Antaeus. This cube of 1,302 records has a dimension called State, with an array of 51 unique values. We'll focus on the three values, "CA" (points to 70 records), "IL" (49 records) and "MA" (56 records). If we set these values so that their records are associated with the colors black, blue and green, respectively, letting all other records be associated with the color gray, any of the cube's 1,024 basic scatter plots will look something this:
Highlighting values in this manner can be very informative. Separation can be applied to any dimension independently, all managed in the Separation SV (SynchroView).
Antaeus also lets you use the values of a dimension to define a subset of records within the cube. This subset can be applied to scatter plots as a brush, which is another, more specific, kind of separation. This cube also has a dimension called Type, which defines the records associated with "Public" and "Private" schools. Now consider the effects of creating a subset from the same State values used above and applying them to the same plot as a brush (yellow), while separating "Public" and "Private" using red and blue, respectively:
And this is just scratching the surface of how dimensions can be used to ask new questions about your data.