Proceedings of the EuSpRIG 2017 Conference “Spreadsheet Risk Management” ISBN : 978-1-905404-54-4
Copyright © 2017, EuSpRIG European Spreadsheet Risks Interest Group (www.eusprig.org) & the Author(s)
While the Formula Diagram gives a global view of the model, the corresponding Formula List gives a
detailed view, with all the formulas written in an Excel-like form, using variable names.
The Formula Diagram is inspired from the Influence Diagram, as presented in (Bodily, 1985). The
Influence Diagram has a richer set of modeling concepts, such as uncertainty in the values of data
variables and uncertainty in the formulas of calculated variables. But the Influence Diagram has no
representation of groups of repeating variables, which the Formula Diagram represents with a dash-
bordered box.
4 Multidimensional modelling concepts
At this point, we invite the reader to read the case study presented in the appendix so that they can get
a better appreciation of the concepts we present in this section.
4.1 Dimensions
A dimension is a set of values that serve to characterize a specific value. The set of values form a
partition. A partition, in set theory, represent subsets whose intersections, taken two by two, are null,
and whose union is the universal set, that is the set of all values. In plain language, it means that there
is no overlap and all possibilities are covered.
For example, if we use the dimension Region to characterize clients and we have the set of values
{Mountain, Valley, Lake}, a client must belong to one of the regions (all possibilities are covered)
and cannot belong to two regions (no overlap).
4.2 Dimension sets
A dimension set is a set comprised of 0 or more dimension, and a variable belongs to a specific
dimension set. Often, the variable name we use gives a clue to the dimension set it belongs to: the
variable named Monthly Production belongs to the dimension set (Month) and the variable Monthly
Regional Sales belongs to the dimension set (Month, Region).
We will say that a variable belonging to the empty, (), dimension set is dimensionless. We will also
say that dimension sets composed of only one dimension are basic. Finally, the dimension set
composed of all the dimensions is called the full dimension set.
If we have ! dimensions, then we have "
#
possible dimension sets, ranging from the empty set to the
set of all dimensions. Thus, when we have only one dimension, like Time, a variable either belongs to
the (Time) dimension set or is dimensionless. If we have two dimensions, like Month and Region, a
variable either belongs to the (Month, Region) dimension set, the (Month) dimension set, the (Region)
dimension set or the () dimension set.
4.3 Defining variables
In usual mathematical notation, a variable’s dimension set is indicated by subscripts. Thus, the two
variables described above would be written like this:
$%&'()*+,-%./0'1%&
$%&'(
and
$%&'()*+2341%&5)+65)37
89:;<=>?@A9:
. It is redundant to specify the dimension set in the variable’s
name and in the subscripts: we will only do so in this section because we want to make sure that the
dimensions are clear.
There are mathematical rules to remember when dealing with expressions involving variables of
different dimension sets. We usually apply them without thinking about it because they are common
sense. We will describe the rules and show how they are represented in a Formula Diagram and a
Formula List.
Rule 1: The dimension set of a formula is the union of the dimension sets of all the variables that are
part of its definition.
Example 1:
• Unit Production Cost is of dimension set (Product).
• Unit Delivery Cost is of dimension set (Region).