Hi @ChrMavrom,
I’m hoping that this might assist. I don’t know of a trivial method of creating a cube dataset in KNIME, so I went for a non-trivial method and (hopefully) built one. 
I have uploaded a demo workflow to the hub, using a copy of your test dataset which I cleaned (removing errant spaces in the type column) and stored in the demo dataflow as a csv.
It makes use of a component I’ve created for the purpose, and is hopefully therefore reusable (but I advise some thorough testing to make sure it suits your needs!!)
The component allows you to select the grouping/dimension columns from your dataset and uses these to build a cube.
Data Input:
Cube Data Output:
Every row is assigned a CubeGroup which is the grouping of the columns represented in that line
You can then use this “CubeGroup” in a GroupBy aggregation along with the other columns you wish to group by (which should all be columns that were supplied to the Cube Builder as dimension columns.
e.g. to get a UNIQUE count of Customer and Order IDs:
The result is a cube summarised by the different levels
And you can then filter / concatenate different CubeGroups …
… to build the required outputs:
From looking at the result, I see that in the test data, some different customers share the same order id, but… you get the idea