[Next] [Previous] [Up] [Top] [Contents]

Example 2 - Extending the `El Farol Bar' Model

General Results


Graphs of the results do not show anything surprising. The average fitnesses of the agents populations of models fluctuate wildly at the beginning with a big difference between agents but as the simulation progresses they all settle down to around the same value (figure 5).



Figure 5. Average fitnesses of the agents' populations of models

The deviance between the fitnesses of the agents' population of models also reduces somewhat but continues to fluctuate at significant levels (figure 6).



Figure 6. Change in variance (in standard deviations) of the Agents' population of models

The graph of the utility of the agents gained over time shows that different agents are doing best at different periods of time, but none for very long - also that the fortunes of all go up and down somewhat together (figure 7).



Figure 7. Utility gained by agents over time (smoothed)

When you look at the pattern of who goes and who does not, some of the agents settle for a fixed strategy but some are more dynamic and constantly swap between different strategies and elaborate old ones (figure 8).



Figure 8. Who did what each week


Modelling Socially Intelligent Agents - Bruce Edmonds - 17 DEC 97
[Next] [Previous] [Up] [Top] [Contents]

Generated with CERN WebMaker