Dear Sirs and Madams,
at the moment im working on a dissertation concerning the following
topic :"Modelling the product-life-cycle of an automobil with System
Dynamics (focus: customer-demand)
I would be very grateful, if anyone had already information about this
topic.
Im also looking forward for any comments and hints concerning the
possible problems which are linked with that topic.
Except that, the most urgent question is: Are there any
"common-solutions" concerning the question how to validate the
causal-chart (and the whole SD-model)? How can you prove that the links
i.e the impacts/effects within the causal-diagramm are "correct" and
reflect the "real-world"?
Any material (models, links, pdfs, book-references, etc) which you think
might help me is welcome explicitly.
Thank you very much indeed for any support!
Torsten Albrecht
--------------------------------------------------------------
Torsten Albrecht
Hegelstrasse 51
96052 Bamberg
Germany
email: <mailto:torsten.albrecht.de@web.de> torsten.albrecht.de@web.de
Validation of SD-models?
-
zenabraham@aol.com
- Junior Member
- Posts: 18
- Joined: Fri Mar 29, 2002 3:39 am
Validation of SD-models?
Greetings,
I dont know how youve designed your model, but I would suggest backcasting.
This is where you "set" the model back to a "time" in the past, before the
intended period that the models run is to cover. Then, you run it "forward"
and evaluate its output results against reality.
I prefer building systems models that are based on profit and loss and
balance sheets of organizations, and demographic and sales data, and then
"mating" that information to submodels reflecting the real-world phenomena
that produced the fiscal numbers.
This attempt is more useful to the people running the organizations you are
attempting to represent, in the first place. It also forces less abstraction
and a harder focus on institutional dynamics.
Im using this practice with a new sim called The Oakland As Microworld.
Its designed around real information on the team, and permits the user to
"run" the baseball organization over a ten-year period. Its not -- yet --
based around players, and when that submodel is installed, one will be asked
to "pitch" a contract the the "player" and a salary structure will result
from that "interaction."
For the present, one can set single game and season ticket prices, total
player payroll, scouting and player development expenses, team marketing
expenses, and build a new stadium. If one chooses to construct a new
stadium, there are nine different revenue streams that can be used in various
combinations. If too much public tax money is proposed for use, the stadium
may not be approved for construction.
In this sim, one cant just charge any ticket price: the higher the price,
the fewer people who will purchase them, depending on team won and loss
record and the economys conditions.
In this, information to model from varies, but the direction of change is the
same. And thats the point. Pay attention to the direction of change in one
variable with respect to change in the variable its dependent on. The
change should be within historical real world perameters.
The objective is to make a sim that feels like the real world youre modeling.
Zennie Abraham
Subject: REPLY Validation of SD-models? (SD3907)
I dont know how youve designed your model, but I would suggest backcasting.
This is where you "set" the model back to a "time" in the past, before the
intended period that the models run is to cover. Then, you run it "forward"
and evaluate its output results against reality.
I prefer building systems models that are based on profit and loss and
balance sheets of organizations, and demographic and sales data, and then
"mating" that information to submodels reflecting the real-world phenomena
that produced the fiscal numbers.
This attempt is more useful to the people running the organizations you are
attempting to represent, in the first place. It also forces less abstraction
and a harder focus on institutional dynamics.
Im using this practice with a new sim called The Oakland As Microworld.
Its designed around real information on the team, and permits the user to
"run" the baseball organization over a ten-year period. Its not -- yet --
based around players, and when that submodel is installed, one will be asked
to "pitch" a contract the the "player" and a salary structure will result
from that "interaction."
For the present, one can set single game and season ticket prices, total
player payroll, scouting and player development expenses, team marketing
expenses, and build a new stadium. If one chooses to construct a new
stadium, there are nine different revenue streams that can be used in various
combinations. If too much public tax money is proposed for use, the stadium
may not be approved for construction.
In this sim, one cant just charge any ticket price: the higher the price,
the fewer people who will purchase them, depending on team won and loss
record and the economys conditions.
In this, information to model from varies, but the direction of change is the
same. And thats the point. Pay attention to the direction of change in one
variable with respect to change in the variable its dependent on. The
change should be within historical real world perameters.
The objective is to make a sim that feels like the real world youre modeling.
Zennie Abraham
Subject: REPLY Validation of SD-models? (SD3907)
-
Yaman Barlas
- Member
- Posts: 44
- Joined: Fri Mar 29, 2002 3:39 am
Validation of SD-models?
Hello Torsten,
I have done considerable work on model validity. Here are some papers you
may find useful:
"A Dynamic Pattern-oriented Test for Model Validation" (with K. Kanar),
Proceedings of 4th Systems Science European Congress, Valencia, Spain,
Sept. 1999, pp. 269-286
ÒFormal Aspects of Model Validity and Validation in System DynamicsÓ,
System Dynamics Review, Vol.12, no.3, 1996, pp. 183-210.
"An Autocorrelation Test for Output Validation", SIMULATION, Vol.55, No.1,
1990, pp.7-16.
"Philosophical Roots of Model Validation: Two Paradigms" (with Stanley
Carpenter), System Dynamics Review, Vol.6, No.2. 1990, pp.148-166.
"Multiple Tests for Validation of System Dynamics Type of Simulation
Models", European Journal of Operations Research, Vol.42, no.1, 1989, pp.
59-87.
"Tests of Model Behavior That Can Detect Structural Flaws: Demonstration
with Simulation Experiments", in Computer-Based Management of Complex
Systems (P. M. Milling and E. O. K. Zahn, eds.), 1989, pp. 246-254.
I have more working papers if you are interested. Please let me know.
Furthermore, we have developed two validity testing software:
1- BTS to test the BEHAVIOR validity of a model
2- STS to test the STRUCTURE of the model (indirectly).
BTS is user friendly and can be downloaded free for non-commercial use from
our SESDYN web site below. (BTS II, more general and powerful version will
be ready in a couple of weeks).
STS is not user friendly yet. We are currently working on it. The first and
last paper above describe the idea and the methods.
regards and good luck with your work!
Yaman Barlas
From: yaman barlas <ybarlas@boun.edu.tr>
---------------------------------------------------------------------------
Yaman Barlas, Ph.D.
Professor, Industrial Engineering Dept.
Bogazici University,
80815 Bebek, Istanbul, TURKEY
Fax. +90-212-265 1800. Tel. +90-212-358 1540; ext.2073
http://www.ie.boun.edu.tr/~barlas
SESDYN Group: http://www.ie.boun.edu.tr/labs/sesdyn/
I have done considerable work on model validity. Here are some papers you
may find useful:
"A Dynamic Pattern-oriented Test for Model Validation" (with K. Kanar),
Proceedings of 4th Systems Science European Congress, Valencia, Spain,
Sept. 1999, pp. 269-286
ÒFormal Aspects of Model Validity and Validation in System DynamicsÓ,
System Dynamics Review, Vol.12, no.3, 1996, pp. 183-210.
"An Autocorrelation Test for Output Validation", SIMULATION, Vol.55, No.1,
1990, pp.7-16.
"Philosophical Roots of Model Validation: Two Paradigms" (with Stanley
Carpenter), System Dynamics Review, Vol.6, No.2. 1990, pp.148-166.
"Multiple Tests for Validation of System Dynamics Type of Simulation
Models", European Journal of Operations Research, Vol.42, no.1, 1989, pp.
59-87.
"Tests of Model Behavior That Can Detect Structural Flaws: Demonstration
with Simulation Experiments", in Computer-Based Management of Complex
Systems (P. M. Milling and E. O. K. Zahn, eds.), 1989, pp. 246-254.
I have more working papers if you are interested. Please let me know.
Furthermore, we have developed two validity testing software:
1- BTS to test the BEHAVIOR validity of a model
2- STS to test the STRUCTURE of the model (indirectly).
BTS is user friendly and can be downloaded free for non-commercial use from
our SESDYN web site below. (BTS II, more general and powerful version will
be ready in a couple of weeks).
STS is not user friendly yet. We are currently working on it. The first and
last paper above describe the idea and the methods.
regards and good luck with your work!
Yaman Barlas
From: yaman barlas <ybarlas@boun.edu.tr>
---------------------------------------------------------------------------
Yaman Barlas, Ph.D.
Professor, Industrial Engineering Dept.
Bogazici University,
80815 Bebek, Istanbul, TURKEY
Fax. +90-212-265 1800. Tel. +90-212-358 1540; ext.2073
http://www.ie.boun.edu.tr/~barlas
SESDYN Group: http://www.ie.boun.edu.tr/labs/sesdyn/