Interactive Visual Decision Analytics Minitrack
Permanent URI for this collection
There is a substantial need for advances in the ability to address large volumes of disparate data, streams from distributed locations, rapid (often real-time) response, and techniques focused on platforms ranging from mobile to high-performance computing. Gaining maximum benefit from these datasets also demands novel ways of supporting the judgment of skilled human decision-makers, such as information visualization, human-information interaction, and studies of individual and collaborative decision-making and coordination of action. Bridging human understanding and computational analysis is necessary not only for analysts but for policy-makers and operational personnel, making this a broadly interdisciplinary effort. Bridging these areas of investigation is key to the success of this minitrack. We believe communication, in the HICSS setting, will cross-pollinate ideas that will advance the state-of-the- art for this class of problems.
From a computational perspective, the challenges of large-volume data, streaming data, and reduced response time initially appear to be quite distinct. Upon closer inspection, they are often found to be encountering very similar constraint issues from different perspectives. Fundamentally all algorithms trade off space, time, data volume, and accuracy. If we can solve a problem on a desktop today, we will expect to be able to solve it on an iPhone or Blackberry tomorrow. If we solve it with cluster, cloud, or supercomputing today, we expect to solve it locally, tomorrow.
Thus this minitrack seeks to bring together those working in areas of massive datasets, streaming data, rapid model construction, across a range of computational footprints and time constraints. It seeks to define analytical methods and technologies that use interactive visualization to meet challenges posed by data, platforms, and applications for decision making and risk-based decision making:
- Visualization and Analysis of datasets of varying size and complexity from archives and real-time streams
- Collaborative visual analysis and operational coordination within and across organizations.
- Interactive and Visual Risk-based decision making
- Interactive Machine Learning methods
- Cross-platform interoperability, from mobiles to data walls
- Managing response time of complex analytical tasks
- Effective deployment and case studies of success from deployed visualization and analytics experiences
- Visualization and analytics for data-driven policy making and decision support
- Issues and Challenges of evaluation of visual decision making
- Cognitive and social science aspects of visual decision making environments
David S. Ebert (Primary Contact)
Simon Fraser University
University of Texas
ItemSpace-Time Kernel Density Estimation for Real-Time Interactive Visual Analytics( 2017-01-04)We present a GPU-based implementation of the Space-Time Kernel Density Estimation (STKDE) that provides massive speed up in analyzing spatial- temporal data. In our work we are able to achieve sub- second performance for data sizes transferable over the Internet in realistic time. We have integrated this into web-based visual interactive analytics tools for analyzing spatial-temporal data. The resulting inte- grated visual analytics (VA) system permits new anal- yses of spatial-temporal data from a variety of sources. Novel, interlinked interface elements permit efficient, meaningful analyses.
ItemHotSketch: Drawing Police Patrol Routes among Spatiotemporal Crime Hotspots( 2017-01-04)During the course of a day, a police unit is expected to move throughout the city to provide a visible presence and respond quickly to emergencies. Planning this movement at the beginning of the shift can provide a helpful first step in ensuring that officers are present in areas of high crime, but these plans can quickly break down as they are pulled away to 911 calls. Once such an initial plan is deferred, police units need to be able to rapidly and fluidly decide where to go next depending on their immediate location and time. In this paper, we present our research to couple spatiotemporal analysis of historical crime data with sketch-based interaction methods. This research is presented through an initial prototype, HotSketch, which we describe through a set of use cases within the domain of police patrol route planning.