Automated parameter specification in dynamic feedback models based on behavior pattern features

Abstract

In various stages of a modeling study, the developer/user encounters the need for identifi cation of specifi c values for certain model parameters. However, due to the non-linear nature of these models it is generally hard to foresee changes in the dynamic behavior as a consequence of even marginal parameter changes. Thus, the need for an automated and effi cient search approach is very clear. A behavior pattern-based parameter search approach and results obtained from various test experiments are presented in this paper. The pattern identifi cation algorithm used enables a parameter search process based on qualitative features of a desired behavior pattern, even in the absence of a reference data series. The approach conducts the search process so as to obtain values for the model parameters that yield an output that matches the desired behavior in terms of its qualitative pattern features. Experiments conducted reveal the viability of the approach as a support tool that can be used in model calibration, sensitivity analysis, validation, and policy design stages.

Using System Dynamics for Grand Challenges: The ESDMA Approach

“…That is, we start by positioning each time series in its own cluster and then hierarchically merging each cluster into larger and larger clusters (Liao, ). Similarity of dynamics is determined from an extension of the behavior pattern features discussed by Yücel and Barlas () (see Yücel, , for details).…”

From Building a Model to Adaptive Robust Decision Making Using Systems Modeling

“…Similar to the development of formal model analysis techniques that smartened the traditional SD approach, new methods, techniques, and tools are currently being developed to smarten modeling and simulation approaches that rely on "brute force" sampling, for example, adaptive output-oriented sampling to span the space of possible dynamics (Islam and Pruyt 2014) or smarter machine learning techniques Kwakkel et al 2014;Pruyt et al 2014c) and time series classification techniques (Yücel and Barlas 2011;Yücel 2012;Sucullu and Yücel 2014;Islam and Pruyt 2014), and (multi-objective) robust optimization techniques .…”

“…Examples of recent developments in simulation setup and execution include model calibration and bootstrapping (Oliva 2003;Dogan 2007), different types of sampling (Fiddaman 2002;Ford 1990;Clemson et al 1995;Islam and Pruyt 2014), multi-model and multimethod simulation Moorlag 2014), and different types of optimization approaches used for a variety of purposes (Coyle 1985;Miller 1998;Coyle 1999;Graham and Ariza 1998;Hamarat et al 2013Hamarat et al , 2014. Recent innovations in model testing, analysis, and visualization of model outputs in SD include the development and application of new methods for sensitivity and uncertainty analysis (Hearne 2010;Eker et al 2014), formal model analysis methods to study the link between structure and behavior Oliva 2008, 2009;Saleh et al 2010), methods for testing policy robustness across wide ranges of uncertainties (Lempert et al 2003), statistical packages and screening techniques (Ford and Flynn 2005;Taylor et al 2010), pattern testing and time series classification techniques (Yücel and Barlas 2011;Yücel 2012;Sucullu andYücel 2014;Islam and Pruyt 2014), and machine learning techniques Kwakkel et al 2014;Pruyt et al 2014c). These methods and techniques can be used together with SD models to identify root causes of problems, to identify adaptive policies that properly address these root causes, to test and optimize the effectiveness of policies across wide ranges of assumptions (i.e., policy robustness), etc.…”

Optimisation Methods in Sustainable Manufacturing

“…Questions regularly arise concerning whether a given policy can be improved, or even what a "good" policy "actually constitutes or entails. In this context, the need for efficient computational methods for policy analysis as well as policy improvement and design has been recognized in system dynamics, see, e.g., Yücel and Barlas (2011), Keloharju and Wolstenholme (1988), and is an active field of research.…”