Data Cubes: operations on cubes

Slicing

Slicing refers to selecting a subset of the cube by choosing a single value for all but two of its dimensions. The result is that you create a slice of the cube, where only two dimensions remain “free” and the data can be visualized as a table.

Going back to the data cube for the Olympics that we used as an example (shown on the left below), if we limit ourselves to only those three dimensions that are shown, some of the ways in which this cube could be sliced are illustrated here:

Dicing

Dicing refers to selecting a subset of the cube by choosing two or more values for multiple dimensions of the cube. The resulting cube has the same number of dimensions, but contains a smaller set of data than the original cube.

The illustration below shows our original Olympics data cube on the left, diced along the following selection of values on the right:

Roll-up

A roll-up involves summarizing the data along a dimension. The summarization rule might be computing totals along a hierarchy, or applying a set of formulas such as "profit = sales - expenses".

In our Olympics data cube, some examples of roll-up are:

Drill-down

Drill-down refers to the exploration of more detailed data, starting from summary data at a higher level in the hierarchy.

Again, in our Olympics data cube, some examples of drill-down are:

Pivoting

When we rotate a data cube in space, this is called pivoting. Pivoting allows us to get another perspective on the data.

Looking at our example cube, if we look at the side of the cube that shows the three types of medals for the three countries for 2012, we have the countries shown as rows, and the medals as columns. Pivoting the cube 90 degree clockwise gives us the same data, but now the countries are shown as columns, and medals as rows. If we pivot that towards us by 90 degrees, countries are still shown as columns, but the rows now show the years.

The underlying data remain unchanged. Pivoting just allows us to see them in different ways.