Rockwell Automation Arena Contact Center Edition Users Guide - Manual (Page 25)

2 • I NTRODUCTION TO S IMULATION
17
•
•
•
• •
2
• In
tr
od
uc
tion
to
Sim
ula
tio
n
enables the modeler to enter them, as well as separate documentation of model assump-
tions, model inputs, and logical relationships.
Validation is the process of raising to an acceptable level the user’s confidence that any
simulation-derived inference about the system is correct. Validation is concerned with
three basic questions:
„
Does the model adequately represent the real-world system?
„
Are the model-generated behavioral data characteristic of the real system’s behavioral
data?
„
Does the simulation model user have confidence in the model’s results?
Consequently, we are concerned with tests that fall into three groups: tests of model
structure, tests of model behavior, and tests of the policy implications of the model.
Because a model is constructed for a specific purpose, its adequacy or validity can only be
evaluated in terms of that purpose. We try to build a model that creates the same problems
and behavioral characteristics as the process or system being studied. Validation occurs
throughout model development, beginning with the start of the study and continuing as
the model builder accumulates confidence that the model behaves plausibly and generates
symptoms or modes of behavior seen in the real system. Validation then expands to
include persons not directly involved in constructing the model.
Validation is a communication process requiring the model builder to communicate the
basis for confidence in a model to a target audience. Unless that confidence can be trans-
ferred, the model’s usefulness will never be realized. Thus, through verification testing,
we develop personal confidence in the model and, through validation measures, transfer
that confidence to others.
We must realize that there are degrees of validation; it is not merely an either-or notion.
Validation is not a binary decision variable indicating whether the model is valid or
invalid. No one or two tests can validate a simulation model. Rather, confidence in the
usefulness of a model must gradually accumulate as the model passes more tests and as
new points of correspondence between model and reality are found. Validation testing
occurs continually in the process of designing, constructing, and using the model.
We should also remember that verification and validation are never really finished. If the
model is to be used for any period of time, the data and the model itself will need periodic
review to ensure validity. Verification and validation are intertwined and proceed
throughout the study. They are not tacked on toward the end of the study; rather, they are
an integral process that starts at the beginning of the study and continues through model
building and model use. It should also be pointed out that involving the ultimate user in
the entire simulation process makes validation much easier.