PEER-OLAF SIEBERS wrote:
>>But no one is actually doing this final step. Or is my impression wrong?
>>Are there any applications (that use MAS and empirical data) that are
>>actually capable of predicting human behaviour?
Among MAS models that have purely positive (i.e., descriptive) motivations,
as opposed to normative/policy-oriented ones, there are a variety that make
use of empirical data. Most commonly, however, this is not data from the
individual level but rather aggregate data. For example, there are a number
of MAS financial market models (e.g., Lux, LeBaron) that do a good job
reproducing the main statistical features of price dynamics in real markets
(e.g., equity, foreign exchange). My work on firm dynamics has the same
character--individual level model used to explain the statistical
properties of firm populations (e.g., size distribution, growth rate
fluctuations); see paper #3 at www.brookings.edu/dynamics/papers.
MAS models that utilize individual-level data are less common, almost
certainly due to the non-availability or difficulty of acquiring such data.
Some of the work in artificial agent anthropology and archaeology does make
use of data on individuals or at least individual 'households'. For
example, our MAS model of the Anasazi uses data on the spatial location of
individual settlements as a basis for comparison of the model output
(Axtell et al, "Population Growth and Collapse in a Multi-Agent Model of
the Kayenta Anasazi in Long House Valley," Proceedings of the National
Academy of Sciences, volume 99, supplement 3, pp. 7275-7279, May 14, 2002).
But, we have only indirect behavioral data on the real Anasazi, of course.
For a general discussion of the distinction between validating one's model
with aggregate vs. individual level data see Axtell and Epstein, "Agent-
Based Modeling: Understanding Our Creations," Bulletin of the Santa Fe
Institute, volume 9, number 2, pp. 28-32, 1994.
Rob Axtell
**************************************************
Center on Social and Economic Dynamics
The Brookings Institution
1775 Massachusetts Avenue, NW
Washington, DC 20036 USA
Phone: (202) 797-6020
Assistant: (202) 797-6093
Fax: (202) 797-2973
http://www.brookings.edu/scholars/raxtell.htm
http://www.brookings.edu/dynamics
|