In modeling the operation of a universitie's library for the purpose of optimization, a common value system would help compare trade-offs.
When designing system models in commercial and industrial sectors, all system variables that are addressed in a 'cost' function for optimization are often converted to a monetary value. A common value system simplifies many aspects of the optimization process.
The values in the academic environment are not driven by commercial concerns so conversion to financial capital does not carry the advantage as in the commercial world.
So I am wondering how to relate the different values to some common currency for comparison and more directly, to a single valued cost function.
Ray
Capital for Optimization
Hi Ray
The problem that you address can exist in a commercial environment too and there are
some things that exist physically and that have no direct system of valuation.
There are different methods to value things.
One classical method when one wants to valuate assets, is to imagine suppressing them
and try to replace them by something else that can be valued, for instance buying the
service delivered by the asset from an external source.
One can get then different possible values.
When one has constructed an interval of possible minimum and maximum value,
the best thing is to make a model where the value is weighted by a parameter
that will make the value vary between the extremes and experiment different policies
adapted to the different values of the parameter. One can then build policies that are
robust and resist best against the full range of variation of the parameter and the interval
of variation of the value.
Regards.
JJ
The problem that you address can exist in a commercial environment too and there are
some things that exist physically and that have no direct system of valuation.
There are different methods to value things.
One classical method when one wants to valuate assets, is to imagine suppressing them
and try to replace them by something else that can be valued, for instance buying the
service delivered by the asset from an external source.
One can get then different possible values.
When one has constructed an interval of possible minimum and maximum value,
the best thing is to make a model where the value is weighted by a parameter
that will make the value vary between the extremes and experiment different policies
adapted to the different values of the parameter. One can then build policies that are
robust and resist best against the full range of variation of the parameter and the interval
of variation of the value.
Regards.
JJ
Capital for Optimization
My thoughts. Tell me if I'm way off on some of my posts because I've been studying SD on my own, and still have lots of questions. The biggest lesson that I've learned is that there is no single right way.
Yes, businesses do like to show how values relate to financials. With a library, the primary focus is what? Literacy? Accessability?
I guess you could have everything leading to one converter labled "values" and that leading to flows of central values. You could show financial values leading to the things purchased, and then those purchases leading other values, represented as fractions of 1. For example, if "accessability" is a value, then complete "accessability" is "1" for "accessability". The values are subjective but valueable to observe, and I think can only be compared relatively, not exactly. It's probably bad form to place too much reliance on exact numbers.
I imagine that you could have one converter for "values" and then lead into it with other values and lead into those values with financial values, and then show how all of these values effect the outputs and outcomes of what the library is supposed to produce.
If "$ for books" leads to ".9" for "accessability", you could also lead into "accessability" with other values that lead to "accessability", like "politeness of librarians". OK, so 1 = "complete politeness". etc...show pros and cons of these things.
Yes, businesses do like to show how values relate to financials. With a library, the primary focus is what? Literacy? Accessability?
I guess you could have everything leading to one converter labled "values" and that leading to flows of central values. You could show financial values leading to the things purchased, and then those purchases leading other values, represented as fractions of 1. For example, if "accessability" is a value, then complete "accessability" is "1" for "accessability". The values are subjective but valueable to observe, and I think can only be compared relatively, not exactly. It's probably bad form to place too much reliance on exact numbers.
I imagine that you could have one converter for "values" and then lead into it with other values and lead into those values with financial values, and then show how all of these values effect the outputs and outcomes of what the library is supposed to produce.
If "$ for books" leads to ".9" for "accessability", you could also lead into "accessability" with other values that lead to "accessability", like "politeness of librarians". OK, so 1 = "complete politeness". etc...show pros and cons of these things.
Hi Ray
The important thing is not how you value things.
Even in commercial matters, you have to evaluate things very approximately.
You have to choose a common unit say utility and value eveything relatively to
that utility measure. The important thing is to see the dynamic of your system and
it will roughly stay the same even if you change the method of valuation.
If you are not convinced you can make the utility values change and see the impact on the
behaviour. You must start with something simple and experiment with your model and try
to understand how it behaves, changing eventually the parameters.
If there is no dymanic to be studied, then your problem is not really a dynamic one, and
needs static methods. It is maybe an allocation problem or something else.
Regards.
JJ
The important thing is not how you value things.
Even in commercial matters, you have to evaluate things very approximately.
You have to choose a common unit say utility and value eveything relatively to
that utility measure. The important thing is to see the dynamic of your system and
it will roughly stay the same even if you change the method of valuation.
If you are not convinced you can make the utility values change and see the impact on the
behaviour. You must start with something simple and experiment with your model and try
to understand how it behaves, changing eventually the parameters.
If there is no dymanic to be studied, then your problem is not really a dynamic one, and
needs static methods. It is maybe an allocation problem or something else.
Regards.
JJ