Concepts, methods, and tools for sensitivity analysis and validation of Complex Adaptive Systems

Complex Adaptive Systems (CASs) are considered vital for addressing major scientific challenges of the 21st century. CASs often defy classical approaches for sensitivity analysis and validation, since they have unconventional properties, like sensitivity to initial conditions, memory behaviour, and adaptability of agents. The resulting lack of insight in the sensitivity and validity of CASs in practice strongly hampers their utility and acceptance.

The present project is meant to remedy this by:

  • Developing new methodologies for sensitivity analysis and validation of CASs;
  • Disseminating the gained knowledge to the CAS-community within WUR.

The strategy in this project is to start with developing and analysing models for minimal CASs relevant to this IPOP theme. The intended CASs are small (2 agents, 1 rule, …), but contain all the essential elements, like agents, adaptability, decision rules, and time evolution. Next, complexity and adaptability are gradually increased by adding agents, organizational levels, decision rules, etc. By systematically analysing the behaviour of these gradually increasing CASs we gain insight in:

  • Effective methodologies for sensitivity analysis and validation of CASs;
  • Necessary requirements for CAS data to be used in sensitivity analysis and validation.

For utility of the methodologies we focus on typical CASs with which we have experience and for which we have data available, among those models on opinion dynamics, land-use change, and agents in business networks.

The gained experience will be disseminated via cooperation with other projects within the IPOP CAS theme, and advanced courses.