Analysis and visualization of the output data from complex simulation models gets increasing attention within modelling as well as within big data analysis communities. We invite you to join a thematic session and a complementary workshop on this topic (see the description below) to be held at the International Congress on Environmental Modelling and Software (iEMSs’2016) in Toulouse, France, on July 10-14, 2016. The deadline to submit an abstract is March 31, 2016.
The session aims to bring together modellers and data analysis specialists to exchange experiences and present their work on the analysis and visualization of output data from simulations, for example from agent-based models.
During the complementary workshop we will present a demo of a new tool for visualization, analysis, and workflow management for agent-based models of socio-ecological systems. The workshop will also further discuss the needs of the modelling community with respect to visualization, analysis of the simulations output data, and it workflow management. The ideal output of the workshop and session would be to write a position paper on the topic.
You are welcome to submit an abstract or a full paper (six to eight pages) by March 31, 2016 to both the session and the workshop. Please see the instructions for submission on the iEMSs 2016 website submission page.
Session B3:Methods for visualization and analysis of high-dimensional simulation model outputs
Organizers: Dawn Parker, Gary Polhill, Tatiana Filatova and Ju-Sung Lee
iEMSs 2016 website sessions page
Simulation models developed to analyze the dynamics of socio-ecological systems (SES) generally produce large quantities of output data, which represent sweeps across a large parameter space and multiple stochastic elements. This is especially true if decision-making of human actors and institutions is modelled in an adaptive and heterogeneous manner, as for example in agent-based models. Researchers are then tasked with the complex task of deriving hypotheses and drawing general conclusions through visualization and analysis of these output data. The development of appropriate and efficient visualization and statistical analysis methods to accomplish this task in social simulations, and consequently for much of SES modeling, remains in the adolescent stage, and communication between researchers across disciplines regarding novel approaches remains limited.
This session is designed as a follow-up to the 2014 IEMSS workshop “Analyzing and synthesizing results from complex socio-ecosystem models with high-dimensional input, parameter and output spaces” and the resulting synthesis paper (Lee et al, 2015, http://jasss.soc.surrey.ac.uk/18/4/4.html), which examined methods currently in use by the socio-ecosystem simulation modelling community. The proposed session specifically targets application of novel visualization and analysis methods not historically used by this community, but potentially in use by other simulation modelling communities (e.g. ecological, engineering applications and social science). Presentations by simulation modellers from other communities are particularly welcomed. Potential topics may include but not be limited to linear and non-linear data dimension reduction strategies such as variants of multidimensional scaling (classical, non-metric, etc.), linear and non-linear principal component analysis, locally-linear embeddings, Isomap, self-organizing maps, various forms of clustering (e.g., stochastic neighbor embedding), kriging (for interpolation in the reduced or unreduced space), graph analytic analysis and visualizations, classification and regression trees, random forests, Bayesian networks, and Boolean modelling. We also welcome proposals using other novel metatmodelling, data visualization, and auralisation methods.
Workshop A4: The miracle prototype: A hands-on demo of a new tool for visualization, analysis, and workflow management for agent-based models of socioeconomic systems
Organizers: Dawn Parker, J. Gary Polhill
IEMSS 2016 website workshops page
This workshop follows up on the 2014 IEMSS workshop “Analyzing and synthesizing results from complex socio-ecosystem models with high-dimensional input, parameter and output spaces,” and a related ESSA metadata workshop, whose inputs contributed to the development of the Digging into Data MIRACLE software prototype, which will be demoed in the workshop. MIRACLE allows users to upload, analyze, and share output data from complex simulation models in a cloud-based community environment. Rather than uploading and executing code, output data, output analysis scripts, and model and provenance metadata are uploaded. Users (project participants, external researchers, or even students) can query data, examine existing visualizations and analyses, and run new queries using novel parameter combinations, sharing and commenting on queries. The tool is designed to facilitate efficient sharing of model output and analysis and thereby improve communication and reduce barriers to entry for modellers.