Evaluation of Measurement Techniques for the Validation of Agent-Based Simulations Against Streaming Data

Timothy W. Schoenharl and Greg Madey

Department of Computer Science and Engineering University of Notre Dame Notre Dame, IN 46556  

Abstract. This paper presents a study evaluating the applicability of several different measures to the validation of Agent-Based Modeling simulations against streaming data. We evaluate the various measurements and validation techniques using pedestrian movement simulations used in the WIPER system. The Wireless Integrated Phone-based Emergency Response (WIPER) system is a Dynamic Data-Driven Application System (DDDAS) that uses a stream of cellular network activity to detect, classify and predict crisis events. The WIPER simulation is essential to classification and prediction tasks, as the simulations model human activity, both in movement and cell phone activity, in an attempt to better understand crisis events.