Difference between revisions of "Preferences"
(Created page with "''Not to be confused with: Rave preferences'' '''Preferences''' are a set of scalar values associated with each data set such that: *Each variable in the data set...") |
|||
Line 4: | Line 4: | ||
*Each variable in the [[data set]] has an associated preference value. | *Each variable in the [[data set]] has an associated preference value. | ||
*The preference values of all variables in the [[data set]] sum to 1. | *The preference values of all variables in the [[data set]] sum to 1. | ||
+ | |||
+ | Note that preferences are a property of [[data sets]], not a property of [[analyses]]. | ||
+ | |||
+ | The fact that the preferences sum to 1 makes them act like "weights" in determining how influential each variable is in contributing to some function. | ||
+ | |||
+ | The reason we call these values "preferences" is that they are most often used when forming multi-attribute utility or objective functions for use in [[optimization]] or other decision making activities. These functions are typically formulated so that larger weight values indicate higher decision maker preference for relative optimality in the corresponding variable. | ||
+ | |||
+ | Rave does not directly use the preference values for any built-in function. They are only used if you create a [[functional variable]] that is a function of the preferences by either: | ||
+ | *Loading a user-supplied function that uses the [[three-variable-input format]], in which case the 2nd input will be the preferences. | ||
+ | *Creating an objective function by clicking the "Create New Objective Fcn" button on the Model tab |
Latest revision as of 09:38, 23 August 2013
Not to be confused with: Rave preferences
Preferences are a set of scalar values associated with each data set such that:
- Each variable in the data set has an associated preference value.
- The preference values of all variables in the data set sum to 1.
Note that preferences are a property of data sets, not a property of analyses.
The fact that the preferences sum to 1 makes them act like "weights" in determining how influential each variable is in contributing to some function.
The reason we call these values "preferences" is that they are most often used when forming multi-attribute utility or objective functions for use in optimization or other decision making activities. These functions are typically formulated so that larger weight values indicate higher decision maker preference for relative optimality in the corresponding variable.
Rave does not directly use the preference values for any built-in function. They are only used if you create a functional variable that is a function of the preferences by either:
- Loading a user-supplied function that uses the three-variable-input format, in which case the 2nd input will be the preferences.
- Creating an objective function by clicking the "Create New Objective Fcn" button on the Model tab