Subspace search and visualization to make sense of alternative clusterings in high-dimensional data

Tatu, Andrada; Maas, Fabian; Färber, Ines; Bertini, Enrico; Schreck, Tobias; Seidl, Thomas; Keim, Daniel A.

doi:10.1109/vast.2012.6400488

Cited by 80 publications

(79 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The system was applicable to any subspace clustering approach. In [11], 2D projections of the data in alternative subspaces were applied, to identify complementary, orthogonal or redundant subspaces; again, the approach was applicable to different subspace selection methods. Another system to rely on subspace cluster comparison is VISA [12], which implement a simple glyph alternative to represent and compare subspace clusters.…”

Section: Related Workmentioning

confidence: 99%

Analysis of Patient Groups and Immunization Results Based on Subspace Clustering

Hund

Sturm

Schreck

et al. 2015

Brain Informatics and Health

Self Cite

View full text Add to dashboard Cite

Abstract. Biomedical experts are increasingly confronted with what is often called Big Data, an important subclass of high-dimensional data. High-dimensional data analysis can be helpful in finding relationships between records and dimensions. However, due to data complexity, experts are decreasingly capable of dealing with increasingly complex data. Mapping higher dimensional data to a smaller number of relevant dimensions is a big challenge due to the curse of dimensionality. Irrelevant, redundant, and conflicting dimensions affect the effectiveness and efficiency of analysis. Furthermore, the possible mappings from highto low-dimensional spaces are ambiguous. For example, the similarity between patients may change by considering different combinations of relevant dimensions (subspaces). We show the potential of subspace analysis for the interpretation of high-dimensional medical data. Specifically, we analyze relationships between patients, sets of patient attributes, and outcomes of a vaccination treatment by means of a subspace clustering approach. We present an analysis workflow and discuss future directions for high-dimensional (medical) data analysis and visual exploration.

show abstract

Section: Related Workmentioning

confidence: 99%

Analysis of Patient Groups and Immunization Results Based on Subspace Clustering

Hund

Sturm

Schreck

et al. 2015

Brain Informatics and Health

Self Cite

View full text Add to dashboard Cite

show abstract

“…Further heuristic interestingness filters for Scatter-and Parallel Coordinate plots have been discussed in [31,9] and may narrow down the potentially large search space for high-dimensional data. In [32], an explorative overview of subspaces contained in high-dimensional data based on mutual differences and clustering quality properties was introduced.…”

Section: Interest-driven Data Filtering For Visual Analysismentioning

confidence: 99%

Feedback-driven interactive exploration of large multidimensional data supported by visual classifier

Behrisch

Korkmaz

Shao

et al. 2014

2014 IEEE Conference on Visual Analytics Science and Technology (VAST)

Self Cite

View full text Add to dashboard Cite

Abstract-The extraction of relevant and meaningful information from multivariate or high-dimensional data is a challenging problem. One reason for this is that the number of possible representations, which might contain relevant information, grows exponentially with the amount of data dimensions. Also, not all views from a possibly large view space, are potentially relevant to a given analysis task or user. Focus+Context or Semantic Zoom Interfaces can help to some extent to efficiently search for interesting views or data segments, yet they show scalability problems for very large data sets. Accordingly, users are confronted with the problem of identifying interesting views, yet the manual exploration of the entire view space becomes ineffective or even infeasible. While certain quality metrics have been proposed recently to identify potentially interesting views, these often are defined in a heuristic way and do not take into account the application or user context. We introduce a framework for a feedback-driven view exploration, inspired by relevance feedback approaches used in Information Retrieval. Our basic idea is that users iteratively express their notion of interestingness when presented with candidate views. From that expression, a model representing the user's preferences, is trained and used to recommend further interesting view candidates. A decision support system monitors the exploration process and assesses the relevance-driven search process for convergence and stability. We present an instantiation of our framework for exploration of Scatter Plot Spaces based on visual features. We demonstrate the effectiveness of this implementation by a case study on two real-world datasets. We also discuss our framework in light of design alternatives and point out its usefulness for development of user-and context-dependent visual exploration systems.

show abstract

“…Often explored direction is finding lower-dimensional representation of the data, which could then be more easily visualized using the standard visualization tools. In [22] and [23], the authors propose methods that explore interactions between examples in subspaces of the original high-dimensional space, and plot these lowerdimensional representations in a form of similarity matrices or scatter plots in order to gain better understanding of the data. However, the methods become intractable as number of examples and dimensions grows, and may not be suitable for large-scale visualization tasks.…”

Section: A Data Visualizationmentioning

confidence: 99%

“…For example, given L = [17,19,16,18,5,7,6], obtained after solving TSP defined by the previous current node 4, we need to solve the TSP defined by the new current node 5. The TSP being solved includes nodes {18, 10, 22, 23, 7}, and the resulting tour is [18,23,10,22,7]. Before moving on to node 7, we replace node 5 with its ordered children to obtain the updated list L = [17,19,16,18,23,10,22,7,6], see Fig.…”

Section: Algorithmmentioning

confidence: 99%

See 1 more Smart Citation

Efficient Visualization of Large-Scale Data Tables through Reordering and Entropy Minimization

Djuric

Vučetić

2013

2013 IEEE 13th International Conference on Data Mining

View full text Add to dashboard Cite

Abstract-Visualization of data tables with n examples and m columns using heatmaps provides a holistic view of the original data. As there are n! ways to order rows and m! ways to order columns, and data tables are typically ordered without regard to visual inspection, heatmaps of the original data tables often appear as noisy images. However, if rows and columns of a data table are ordered such that similar rows and similar columns are grouped together, a heatmap may provide a deep insight into the underlying data distribution. We propose an informationtheoretic approach to produce a well-ordered data table. In particular, we search for ordering that minimizes entropy of residuals of predictive coding applied on the ordered data table. This formalization leads to a novel ordering procedure, EMordering, that can be applied separately on rows and columns. For ordering of rows, EM-ordering repeats until convergence the steps of (1) rescaling columns and (2) solving a Traveling Salesman Problem (TSP) where rows are treated as cities. To allow fast ordering of large data tables, we propose an efficient TSP heuristic with modest O n log(n) time complexity. When compared to the existing state-of-the-art reordering approaches, we show that the method often provides heatmaps of higher visual quality, while being significantly more scalable. Moreover, analysis of real-world traffic and financial data sets using the proposed method, which allowed us to readily gain deeper insights about the data, further confirmed that EM-ordering can be a valuable tool for visual exploration of large-scale data sets.

show abstract

Subspace search and visualization to make sense of alternative clusterings in high-dimensional data

Cited by 80 publications

References 24 publications

Analysis of Patient Groups and Immunization Results Based on Subspace Clustering

Analysis of Patient Groups and Immunization Results Based on Subspace Clustering

Feedback-driven interactive exploration of large multidimensional data supported by visual classifier

Efficient Visualization of Large-Scale Data Tables through Reordering and Entropy Minimization

Contact Info

Product

Resources

About