Next Issue: Pitfalls in Implementing
Modeling Solutions
Previous Issue:
Can you prevent loosing customers?
Standard graphs such as pie charts, line
graphs, bubble diagrams, 3D graphs are widely used to describe the relationship
between different variables. However, they fail to capture the effect
of one variable on the other. It is possible to argue that such effects
do in fact exist  as mental pictures. In which case, to create an effective
diagram, the skill is to put the mental picture onto paper so that the
reader obtains exactly the same information.
Reference Diagrams come handy at such situations.
Standard graphs often create assumptions and doesn't provide a clear picture
of the situation. On the other hand reference diagrams are simpler
to understand. They present straightforward
scenarios even if there are changes in relationships or sequences.
These diagrams are used to express complex relationships
in a graphical manner so that change in one variable in the whole diagram
shows where all its going to impact and to what extent.
A typical Reference Diagram has following
components 
 Object: The objects interacting with each other in the system.
 Relation/Association: A link connecting the associated objects.
Qualifiers can be placed on either end of the association to depict
cardinality.
 Messages: The interaction between different objects is represented
as messages.
These diagrams are very useful tools to understand
the flow of the model. For example, a model with an objective function
of minimizing logistics and inventory carrying cost subject to various
feasibility and service level constraints can be represented through a
reference diagram.
What Reference Diagrams contribute to build
a model:
 The context: A model is typically a set of mathematical equations
and these are difficult to understand by a normal user. Reference Diagram
graphically clarifies the model context.
 What something does: Generally model reformats the data and
finds some hidden patterns, which are easier to understand with the
diagrams. A diagram can show the structure of the input data and the
output data. A typical example for such a scenario would be a Customer
Order (input data) received gets translated into a Sales Order (output
data).
 Detail overview: Model consists of constraints, preferences,
and sequences apart from input and output data. All these details are
easier to understand and explain with the use of reference diagrams.
 Components: During a debug stage, models are hard to debug
so reference models are useful tool to define components of the model
and allow user to debug the model component wise.
 Comparisons: It is easier to compare different models with
the use of reference diagrams.
 Relationships: A diagram that shows components is also likely
to show the relationships between the components. All preprocessing
or postprocessing components of the model are easier to relate to inputs
and output.
 Sequences: This is a common use of diagrams and it enables
authors to show the correct order for a sequence of events or tasks.
It's worth mentioning that reference diagrams convey more than one type
of information. Reference diagrams convert complexity of model to simple
and useful context.
©
200010 DecisionCraft Inc.
