Dr. Andrada Tatu
|Telefon:||(+49) 07531 88-4364|
|Fax:||(+49) 07531 88-3268|
|Adresse:||Universität Konstanz |
Fachbereich Informatik und Informationswissenschaft
78457 Konstanz, Germany
Visual Analytics of Patterns in High-dimensional Data
The extraction of relevant and meaningful information out of high-dimensional data is notoriously complex and cumbersome. The curse of dimensionality is a popular way of stigmatizing the whole set of troubles encountered in high-dimensional data analysis; finding relevant projections, selecting meaningful dimensions, and getting rid of noise, being only a few of them. Multi-dimensional data visualization also carries its own set of challenges like, above all, the limited capability of any technique to scale to more than an handful of data dimensions.
Visual exploration of multivariate data typically requires projection onto lower-dimensional representations. The number of possible representations grows rapidly with the number of dimensions, and manual exploration quickly becomes ineffective or even unfeasible. Visual quality measures have been recently devised to automatically extract interesting visual projections out of a large number of available candidates. The measures permit for instance to search within a large set of scatter plots (e.g., in a scatter plot matrix) and select the views that contain the best separation among clusters. Using quality measures, the user is provided with a manageable number of potentially useful candidate visualizations, which can effectively ease the task of finding truly useful visualizations and speed up the data exploration task. We developed measures for class-based as well as non class-based scatter plots and parallel coordinates visualizations (TVCG '11).
We also provide an overview of approaches that use quality metrics in high-dimensional data visualization and propose a systematization based on a thorough literature review (InfoVis '11).
Interesting patterns may be located in subspaces of a large input feature space.
While a rich body of research has been carried out in designing subspace clustering algorithms, surprisingly little attention has been paid to develop effective visualization tools to help analyzing the clustering result. Appropriate visualization techniques could not only help in monitoring the clustering process but,
they also enable the domain expert to guide and even to steer the subspace clustering process to reveal the patterns of interest. To this goal we envision a concept that combines scalable subspace clustering algorithms and interactive scalable visual exploration techniques. This work will include the task of comparative visualization and feedback guided computation of multiple alternative clusterings.
Projects and Paper information
Subspace Search and Visualization to Make Sense of Alternative Clusterings in High-Dimensional Data
A systematic methodology for visual analysis of high-dimensional data where an interestingness-guided subspace search algorithm detects interesting subspaces. Based on appropriately defined subspace similarity functions, we visualize the subspaces and provide navigation facilities to interactively explore large sets of subspaces. Our approach allows users to effectively compare and relate subspaces with respect to the involved dimensions and clusters of objects.
A Taxonomy on Visual Cluster Separation Factors
Qualitative evaluation of 816 plots, including analysis of the reasons for failure of previous cluster separation metrics and a taxonomy of factors that affect separation.
Quality Metrics in High-Dimensional Data Visualization: An Overview and Systematization
Literature review for quality metrics used in high-dimensional data visualization leading to the definition of a number of factors that characterize this works and a quality metrics pipeline.
|10/2009 - 07/2013||PhD Student|
|2007 - 2009||Master of Science in Information Engineering at the University of Konstanz, Germany|
|2004 - 2007||Bachelor of Science in Information Engineering at the University of Konstanz, Germany|
M. Hund, I. Färber, M. Behrisch, A. Tatu, T. Schreck, D. A. Keim and T. Seidl.
Visual Quality Assessment of Subspace Clusterings.
KDD 2016 Workshop on Interactive Data Exploration and Analytics (IDEA’16), 2016.
Visual Analytics of Patterns in High-Dimensional Data.
M. Schaefer, L. Zhang, T. Schreck, A. Tatu, J. A. Lee, M. Verleysen and D. A. Keim.
Improving projection-based data analysis by feature space transformations.
In Proceedings of VDA 2013, 2013.
A. Tatu, L. Zhang, E. Bertini, T. Schreck, D. A. Keim, S. Bremm and T. von Landesberger.
ClustNails: Visual Analysis of Subspace Clusters.
Tsinghua Science and Technology, Special Issue on Visualization and Computer Graphics, 17(4):419-428, 2012.
M. Sedlmair, A. Tatu, T. Munzner and M. Tory.
A Taxonomy of Visual Cluster Separation Factors.
Computer Graphics Forum (Proc. EuroVis 2012), 31(3):1335-1344, 2012.
E. Bertini, A. Tatu and D. A. Keim.
Quality Metrics in High-Dimensional Data Visualization: An Overview and Systematization.
IEEE Symposium on Information Visualization (InfoVis), 17(12):pages 2203-2212, 2011.
D. J. Lehmann, G. Albuquerque, M. Eisemann, A. Tatu, D. A. Keim, H. Schumann, M. Magnor and H. Theisel.
Visualisierung und Analyse multidimensionaler Datensätze.
Informatik-Spektrum, Springer Berlin / Heidelberg, 33(6):589-600, DOI: 10.1007/s00287-010-0481-z, 2010.