The Visual Analytics Group researches and develops interactive systems for analysis and decision support.
The focus of the group is on multi- and high-dimensional data which is oftentimes time-dependent. Our goal is to help users acquire new insights and an increased decision confidence by enabling an interactive dialog with big structured data. For this purpose, we create novel technologies which combine visualization, user interaction, and automatized analysis methods such as data mining with response in real time. The application domains of current projects comprise, e.g., the energy sector, industrial manufacturing, health policy making, and simulation-based development of powertrain systems.
Our research focuses on creating effective visual techniques for supporting tasks along the process of data analysis:
- Data quality assessment and cleansing of complex data, e.g., by visualizing plausibility checks
- Getting an overview of high-dimensional data and recognizing dependencies in this data
- Building and validating regression models, e.g., predictive models of energy time series as well as surrogate models of time-consuming physical simulations
- Multi-criteria decision making, e.g., for supporting statistical model selection as well as for decisions in engineering
An additional research topic concerns the creation of highly responsive software infrastructures which provide immediate feedback on user interaction even in case of large data. In this context, we focus on tightly integrating popular analytical platforms such as Matlab, R, and Python, as well as on defining flexible modules which permit a rapid prototyping of task-specific analytical applications.
- Information visualization
- Visual Analytics
- Machine Learning
- Analytical software architectures
- Human computer interaction
- Data science for conducting specific data analyses
MÜHLBACHER T. and PIRINGER H.: A Partition-Based Framework for Building and Validating Regression Models. Accepted for publication in IEEE Transactions on Visualization and Computer Graphics (TVCG), vol. 19, num. 12, pp. 1962 – 1971, 2013.
Best paper award at IEEE VAST 2013
BERGER W., PIRINGER H., FILZMOSER P., GRÖLLER E.: Uncertainty-Aware Exploration of Continuous Parameter Spaces Using Multivariate Prediction. Computer Graphics Forum Volume 30, Number 3, pp. 911 – 920, 2011.
Best paper award at EuroVis 2011
MÜHLBACHER T., PIRINGER H., GRATZL S., SEDLMAIR M., and STREIT M.: Opening the Black Box: Strategies for Increased User Involvement in Existing Algorithm Implementations. IEEE Transactions on Visualization and Computer Graphics, vol. 20, iss. 12, pp. 1643 – 1652, 2014
PIRINGER H., PAJER S., BERGER W., TEICHMANN H.: Comparative Visual Analysis of 2D Function Ensembles. Computer Graphics Forum Volume 31, Number 3, pages 1195 – 1204, 2012
SORGER J., ORTNER T., LUKSCH C., SCHWÄRZLER M., GRÖLLER E., and PIRINGER H.: LiteVis: Integrated Visualization for Simulation-Based Decision Support in Lighting Design. IEEE Transactions on Visualization and Computer Graphics (VAST 2015), vol. 22, iss. 1, pp. 290 - 299, 2016
Visual Analytics Software-Platform Visplore
Visplore is a software platform permitting a rapid and flexible creation of task-specific analytical applications interactively by the user. Visplore has been designed for the visual exploration of multi- to high-dimensional data, including categorical and functional data. The focus is on supporting specific tasks such as data quality assessment, process optimization, and the analysis of prediction models. The software architecture exploits parallelism in order to support an interactive analysis of millions of data records in real time. Visplore has served as software platform for numerous scientific results of the Visual Analysis group, as well as for applications, e.g., in the automotive industry, the energy sector, industrial manufacturing, the health sector, and for network security.
- Visuals 3.0
- Visuelle Datenanalyse im Qualitätsmanagement
- A Partition-Based Framework for Building and Validating Regression Models
- Visplore Modellvalidierung
- TSM Visuals
- Weight Lifter
Energy sector (partners: Austrian Power Grid AG, HAKOM Solutions GmbH): Quality assessment of time series data; validation and comparison of predictive models
Powertrain engineering (partner: AVL List GmbH): Validation of simulation data, sensitivity analysis, multi-objective optimization
Industrial manufacturing (partners: RHI AG, Plasmo Industrietechnik GmbH): Monitoring of production processes and production fault; analysis for process optimization
Health economics (partners: Hauptverband österreichischer Sozialversicherungesträger, Gesundheit Österreich GmbH): Quality assessment of accounting data, monitoring and analysis of health system performance indicators
Aviation (partners: Schedule Coordination Austria GmbH, AI-MS Aviation Infrastructure Management Systems GmbH): Identification of correlations and trends, data preparation, detail analysis of flight schedule optimization results
A powerful basis for implementing customised analysis applications.
En2VA develops new methodologies for cleansing, predictive modelling, and presenting high-dimensional data from manufacturing, engineering, and energy.
Decison Support for Health Policy and Planning: Methods, Models and Technologies based on existing health care data.
Task-Oriented Visual Analysis of High-Dimensional Data
Visual Analysis and Rendering: Fundamental research towards the combination and integration of the spatial and abstract domain.
Situation Awareness for Infrastructure Operators
Visualisation-Centred Workflow for Interactive Data Analysis
Visual Exploration of Energy Data
Strategic Research Project
Interactive Visual Analysis.
Task-Oriented Visualization Techniques for the Simulation-Based Optimization of Combustion Engines.