Abstract for: Applications of Calibration Techniques to Dynamic Models

Our System Dynamics conference session will focus on two calibration techniques, specifically Approximate Bayesian Computation (ABC) and Pattern-Oriented Modeling. These techniques are of particular interest to discuss because of their fairly recent application in simulation modeling, as well as robust model results that will be exhibited in accompanying case studies. Increasingly, we are using these techniques in our modeling work at PWC. In addition, there is a growing literature that is developing new techniques for calibrating dynamic models. We will walk through a step-by-step overview of both approaches, when and why to use each approach and caveats. For Approximate Bayesian Computation, we will take a deep dive into recent project for the Clinton Foundation, a non-profit philanthropy, specifically modeling women’s barriers and enablers’ impact on female labor force participation and GDP. For Pattern-Oriented Modeling, I will discuss a recent Auto industry case on entering a newly developing product market with little historical data for calibration.