Simulation is a useful and effective way to analyze and study complex, real-world systems. It allows researchers, practitioners, and decision makers to make sense of the inner working of a system that involves many factors often resulting in some sort of emergent behavior. The number parameter value combinations grows exponentially and quickly becomes infeasible to test them all or even to explore a suitable subset of them. How does one then efficiently identify the parameter value combinations that matter for a particular simulation study? In addition, is it possible to train a machine learning model to predict the outcome of an agent-based model without running the agent-based model for all parameter value combinations? We propose utilizing covering arrays to create \(t\)-way (\(t\) = 2, 3, 4, etc.) combinations of parameter values to significantly reduce the parameter value exploration space for agent-based models. In our prior work we showed that covering arrays were useful for systemat- ically decreasing the parameter space in an agent-based model. We now build on that work by applying it to Wilensky’s HeatBugs model and training a random forest machine learning model to predict simulation results by using the covering arrays to select our training and test data. Our results show that a 2-way covering array provides sufficient training data to train our random forest to predict three different simulation out- comes. Our process of using covering arrays to decrease parameter space to then predict ABM results using machine learning is successful.
Simulation is a useful and effective way to analyze and study complex, real-world systems. It allows researchers, practitioners, and decision makers to make sense of the inner working of a system that involves many factors often resulting in some sort of emergent behavior. The number parameter value...
See full abstract
Simulation is a useful and effective way to analyze and study complex, real-world systems. It allows researchers, practitioners, and decision makers to make sense of the inner working of a system that involves many factors often resulting in some sort of emergent behavior. The number parameter value combinations grows exponentially and quickly becomes infeasible to test them all or even to explore a suitable subset of them. How does one then efficiently identify the parameter value combinations that matter for a particular simulation study? In addition, is it possible to train a machine learning model to predict the outcome of an agent-based model without running the agent-based model for all parameter value combinations? We propose utilizing covering arrays to create \(t\)-way (\(t\) = 2, 3, 4, etc.) combinations of parameter values to significantly reduce the parameter value exploration space for agent-based models. In our prior work we showed that covering arrays were useful for systemat- ically decreasing the parameter space in an agent-based model. We now build on that work by applying it to Wilensky’s HeatBugs model and training a random forest machine learning model to predict simulation results by using the covering arrays to select our training and test data. Our results show that a 2-way covering array provides sufficient training data to train our random forest to predict three different simulation out- comes. Our process of using covering arrays to decrease parameter space to then predict ABM results using machine learning is successful.
Hide full abstract
