According to some authors (i.e. Stach, 2005. "Genetic learning of fuzzy cognitive maps") and my own working experience, FCMs with a large number of nodes (i.e. >15-20) and high (edge) density (80-90%), are hard to comprehend and analyze, which, besides, results in difficulties when it comes to interpreting causal- relationship simulations. What modelling technique, based on fuzzy expert knowledge, would you recommend as an alternative?
I think it depends on the purpose and goal of your modelling. For instance, you can use system dynamics with fuzzy effects between variables for understanding complex phenomena and simulate the system. Our group, for instance, just finished a framework like this for marketing problems. I also recommend you paper from Voinov and Bousquet 2010 called 'modelling with stakeholders' where you can find descriptions of different modelling methodologies.
Regards
@Manuel, thanks for your response and interesting tips. Is there any publication about your marketing case that I could check to make an idea of the methodology you used?
Francisco,
what do you think about Ordered Fuzzy Numbers funded by Witold Kosinski?
(https://www.researchgate.net/profile/Witold_Kosinski)
You will have a chance to catch a trend of changes in this way.
Best regards,
Jacek
- Badges
- Science topic