Open map List

Pattern-Oriented Modelling (POM)

is a strategy in agent-based modeling (ABM) that focuses on using multiple patterns observed in real-world data to inform and validate the model. It aims to ensure that the agent-based model reproduces key patterns seen in empirical data, thereby increasing the model's reliability and credibility.

Introduction

POM promotes modelling system accross scales and multiple level. It applied hierarchie of complex system.

POM can be used as a strategy for:

  • multicriteria design
  • selection and calibration/validation of models

Explanation

Three steps (elements) of POM

  • define a structure of your model, using the model purpose as a filter
  • identify the pattern that characterize the system
  • define criteria for deciding whether you reproduced the patterns
  • refive the model structure (iterate over steps 1-3)

 

Pattern in POM are used to:

  • determine scale, entities, variable and processes 
  • test and select sub models
  • find parameter values during calibration

 

Hypothesis in POM

  • identify alternative submodels that implement alternative hypothesis
  • implement the submodels
  • contract the alternatives
  • rrepeat until a submodel has been found that reproduces

 

Examples

in ecosystem symulation:

Spatial distribution of species, population dynamics over time, interaction networks between species.

Outgoing relations

Incoming relations