原创力文档- 1、本文档共24页，可阅读全部内容。
- 2、原创力文档（book118）网站文档一经付费（服务费），不意味着购买了该文档的版权，仅供个人/单位学习、研究之用，不得用于商业用途，未经授权，严禁复制、发行、汇编、翻译或者网络传播等，侵权必究。
- 3、本站所有内容均由合作方或网友上传，本站不对文档的完整性、权威性及其观点立场正确性做任何保证或承诺！文档内容仅供研究参考，付费前请自行鉴别。如您付费，意味着您自己接受本站规则且自行承担风险，本站不退款、不进行额外附加服务；查看《如何避免下载的几个坑》。如果您已付费下载过本站文档，您可以点击 这里二次下载。
- 4、如文档侵犯商业秘密、侵犯著作权、侵犯人身权等，请点击“版权申诉”（推荐），也可以打举报电话：400-050-0827(电话支持时间：9:00-18:30)。
查看更多
19
Design&Implementationof
Optimization-BasedDecision
SupportSystems
I don't want it perfect, I want it Thursday.
-John Pierpont Morgan
19.1 General Structure of the Modeling Process
The overall modeling process is one of:
1) Determining the need for a model.
2) Developing the model.
3) Implementing the model.
One should not skip over step (1) in one’s enthusiasm to use a fancy model. Some questions to ask
before deciding to use a model are:
a) Are the expected savings from using the model greater than the cost of developing and
implementing the model?
b) Is there sufficient time to do (2) and (3) before the recommendation is needed?
c) Is it easier to do an experiment on the real system than to build a model? Sometime ago a
question arose in a telephone company about the effect of serving certain telephone calls
for information arising ity A from a central facility ity B, rather than from the
existing facility in A. It was more simple to make a five minute wiring change to see
what happened, than to construct an accurate statistical model. Similarly, Banks and
Gibson (1997) describe a situation in which one fast food restaurant chain built a detailed
model to evaluate the effect of a second drive-up window. In less time, a competitor
tested the same idea by stationing a second person in the drive-up lane with a remote
hand-held terminal and voice communication to the inside of the restaurant.
d) Do we have the input data needed to make plausible use of
文档评论（0）