Correlation Judgment and Visualization Feature: A Comparative Study

Recent visualization research efforts have incorporated experimental techniques and perceptual models from the vision science community. Perceptual laws such as Weber's law, for example, have been used to model the perception of correlation in scatterplots. While this thread of research has progressively refined the modeling of the perception of correlation in scatterplots, it remains unclear as to why such perception can be modeled using relatively simple functions, e.g., linear and log-linear. In this paper, we investigate a longstanding hypothesis that people use visual features in a chart as a proxy for statistical measures like correlation. For a given scatterplot, we extract 49 candidate visual features and evaluate which best align with existing models and participant judgments. The results support the hypothesis that people attend to a small number of visual features when discriminating correlation in scatterplots. We discuss how this result may account for prior conflicting findings, and how visual features provide a baseline for future model-based approaches in visualization evaluation and design. @article{yang2018correlation, title={Correlation Judgment and Visualization Features: A Comparative Study}, author={Yang, Fumeng and Harrison, Lane and Rensink, Ronald A and Franconeri, Steven and Chang, Remco}, journal={IEEE Transactions on Visualization and Computer Graphics}, year={2018}, publisher={IEEE}}

Encoding Time Series as Multi-Scale Signed Recurrence Plots for Classification Using Fully Convolutional Networks

Recent advances in time series classification (TSC) have exploited deep neural networks (DNN) to improve the performance. One promising approach encodes time series as recurrence plot (RP) images for the sake of leveraging the state-of-the-art DNN to achieve accuracy. Such an approach has been shown to achieve impressive results, raising the interest of the community in it. However, it remains unsolved how to handle not only the variability in the distinctive region scale and the length of sequences but also the tendency confusion problem. In this paper, we tackle the problem using Multi-scale Signed Recurrence Plots (MS-RP), an improvement of RP, and propose a novel method based on MS-RP images and Fully Convolutional Networks (FCN) for TSC. This method first introduces phase space dimension and time delay embedding of RP to produce multi-scale RP images; then, with the use of asymmetrical structure, constructed RP images can represent very long sequences (>700 points). Next, MS-RP images are obtained by multiplying designed sign masks in order to remove the tendency confusion. Finally, FCN is trained with MS-RP images to perform classification. Experimental results on 45 benchmark datasets demonstrate that our method improves the state-of-the-art in terms of classification accuracy and visualization evaluation.

Open our visualization eyes, individualization: On Albrecht Dürer’s 1515 wood cut celestial charts

The position that visualization is an intimate part of human existence and associated with the human species is advanced in this work: visualization abounds delimited by the space of individuality across human history. Visualization involves two complementary aspects of the uniqueness deemed of individuals: individualization reflects individuals’ capabilities and personalization reflects designs that seek compatibility with individuals’ capabilities. This has a number of implications upon the design and evaluation of visualizations. For one, a suitable visualization model that expresses individualization and personalization is needed: a brief survey of models is presented. For another, addressing intellectual uniqueness requires deep analysis and selective objective balance due to the potentially humongous number of unique ideas that support visualization design and viewer experiences. The Engineering Insightful Serviceable Visualizations model is selected as a guide for a comprehensive visualization evaluation of Albrecht Dürer’s 1515 celestial charts. Motivating this choice of visualization is its significance as the first notable and influential European star chart intended for scientific use and mass viewership, and as a blending of science and art. In addition, there is a lack of discussion concerning this particular visualization in the visualization literature. Concluding remarks suggest the significance of approaching visualization from this point-of-view.

A visual approach to creating multivariate geovisualization test data

<p><strong>Abstract.</strong> Testing multivariate geovisualizations, for example glyph designs, for their perceptual qualities requires suitable test data. Synthetic data can be useful for evaluating different data characteristics and their perceptibility through different designs. Real data may not contain all relevant data characteristics, for example with regard to spatial distributions or trends, which may be interesting for testing. Additionally, real data may contain a lot of noise or randomness. Even though, real data are imperative for testing design performance in realistic situations. Fuchs et al. (2017) did a systematic review on 64 glyph evaluation studies, whereof more than 60% used synthetic data. They note that for a better understanding of glyph designs more studies should evaluate glyphs with synthetic and real data. Besides visualization evaluation, a number of application areas, e.g. machine learning or software testing, generate and use synthetic data. Thus, many methods and tools for generating synthetic data exist. Typically, a mathematical function or statistical distribution is defined and random or ordered values, optionally overlaid with noise, are drawn from the defined models to build the synthetic data. However, we found it difficult to create, especially multivariate, spatial distributions of data that follow specific rules and display interactions between the data dimensions with existing tools. Thus, we designed a process that allows the intuitive ‘drawing’ of spatial data distributions and subsequently the derivation of multivariate data from several overlaid layers of ‘drawings’.</p>

An empirical study of data visualization techniques in PACS design

The paper presents an empirical study of multidimensional visualization techniques. The study is motivated by the problem of decision making in PACS (Picture Archiving and Communications System) design. A comprehensive survey of visualizations used in literature is performed and these survey results are then used to produce the final set of considered visualizations: tables (as control), scatterplots, parallel coordinates, and star plots. An electronic testing tool is developed to present visualizations to three sets of experimental subjects in order to determine which visualization technique allows users to make the correct decision in a sample decision making problem based on real-world data. Statistical analysis of the results demonstrates that visualizations show better results in decision support than tables. Further, when number of dimensions is large, 2D parallel coordinates show the best results in accuracy. The contribution of the presented research operates on two levels of abstraction. On the object level, it provides useful data regarding the relative merits of visualization techniques for the considered narrow use-case, which can then be generalized to other similar problem sets. On the meta level above, it contributes an enhanced methodology to the area of empirical visualization evaluation methods.