An Active Learning Exercise for Introducing Agent-Based Modeling

Recent developments in agent-based modeling as a method of systems analysis and optimization indicate that students in business analytics need an introduction to the terminology, concepts, and framework of agent-based modeling. This article presents an active learning exercise for MBA students in business analytics that demonstrates agent-based modeling by solving a knapsack optimization problem. For the activity, students act as naiive agents by using dice to randomly selecting items for a finite capacity knapsack to maximize the value of the knapsack. Students then design a greedy heuristic to skew the probability of selection item. These pencil-and-paper models are then implemented in a spreadsheet model to demonstrate the effects of altering the agents' behavior. Finally, a binary integer programming model is examined to contrast agent-based modeling with traditional mathematical programming formulations. This exercise is innovative because it combines student engagement via active learning with an innovative, individual-based, modeling methodology.