DiD: MIning Relationships Among variables in large datasets from CompLEx systems (MIRACLE) DiD: MIning Relationships Among variables in large datasets from CompLEx systems (MIRACLE) PROJECT SUMMARY Overview: Page A With advances in high-performance computing and the improved availability of finescale socio-economic data, social science researchers have developed new computational modeling methods to explore the dynamics and consequences of human interactions. These methods include agent-based models (ABMs), which create virtual worlds of social agents and their interaction environments in computer code. Adoption of these new methods has been motivated by a desire to explore how interactions between heterogeneous, adaptive social actors at a fine scale collectively shape emergent social outcomes at aggregate scales in complex socio-ecological systems -- questions that cannot be investigated using traditional equilibrium-based and statistical modeling approaches. The non-equilibrium, computational structure of ABMs allows the creation of models with a sufficiently high level of detail to explore such questions. Our proposed project seeks to address this challenge by developing, applying and disseminating an integrated environment for visualization and analysis of data generated by complex systems. The project builds on existing analysis algorithms from each of our projects: prototype web-based data management, analysis, and communication tools; and the NSF-supported CoMSES Net Computational Modeling Library (CML). Each research team will contribute the output data from their computational modeling efforts, as well as existing and newly developed analysis algorithms, to the CML. . Intellectual Merit : In the last decade, ABMs modelers have made substantial progress to identify compelling research questions, design models to investigate them, and develop methods to bring real-world data into models at both the creation and evaluation stage. Since the ABMs we design have stochastic elements and many potential parameter combinations, multiple model runs that sweep parameters are conducted, creating large quantities of computationally generated, hyper-dimensional, "big data"from which we hope to extract answers to social science research questions. Yet, we lack appropriate methods to mine, analyze, and synthesis large-scale model output data in order to answer our questions. Traditional analysis methods for mapping relationships between input parameters and output data -- in both real-world and computational data -- are designed for data that are linear, continuous, and normally distributed. However, data from models of complex socio-ecological systems are non-linear, discontinuous, and power-law distributed. Our project builds on the expertise and established collaborative relationships of a team of senior and emerging junior researchers. Several team members (Parker, Polhill, and Barton) have been early adopters of these models and leaders in the development of methods, priorities, and infrastructure to support them. Our analysis tools and platform will provide an automated means of discovering relationships among variables offer an opportunity to uncover new theories about how social systems work, test the realism of the simulation against knowledge from empirical systems, and propose new parameter subspaces to explore. . Broader Impacts : Each team will publish the results of their data analysis, comparing application of the algorithms between their generated and real-world validation data. On the CoMSES Net platform, a new suite of collaborative, open-source tools will be developed that will allow any users to 1) post their model output, as well as enter metadata describing the structure of their output data 2) visualize and analyze their output data using our new tools 3) comment on and share their analysis 4) conduct comparative and meta-analysis, drawing on output data from other projects. The collaborative tools will improve communication and reduce barriers to entry to new users. Members of our research team are experienced in the use of ABMs to explore dynamic feedbacks between human resource use and the biophysical environment to develop improved policies for sustainable resource use.
|Effective start/end date||8/1/14 → 7/31/18|
- National Science Foundation (NSF): $124,989.00
Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.