Surfacing Visualization Mirages

McNutt, Andrew; Kindlmann, Gordon; Correll, Michael

doi:10.1145/3313831.3376420

Cited by 70 publications

(75 citation statements)

References 85 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We believe that our type-based template search will dovetail with a metamorphic testing [58] based validation approach: by varying the parameters of a template and comparing the resulting images, a validation system could automatically identify errors at the intersection of data and encoding. Similarly, we suggest that templates likely ofer an opportune medium for applying visualization linting [32,57] to a visual analytics context, as the types expose the specifc arguments over which analysis could be conducted.…”

Section: Validation In Visualmentioning

confidence: 99%

Integrated Visualization Editing via Parameterized Declarative Templates

McNutt

Chugh

2021

Proceedings of the 2021 CHI Conference on Human Factors in Computing Systems

View full text Add to dashboard Cite

Interfaces for creating visualizations typically embrace one of several common forms. Textual specifcation enables fne-grained control, shelf building facilitates rapid exploration, while chart choosing promotes immediacy and simplicity. Ideally these approaches could be unifed to integrate the user-and usage-dependent benefts found in each modality, yet these forms remain distinct.We propose parameterized declarative templates, a simple abstraction mechanism over JSON-based visualization grammars, as a foundation for multimodal visualization editors. We demonstrate how templates can facilitate organization and reuse by factoring the more than 160 charts that constitute Vega-Lite's example gallery into approximately 40 templates. We exemplify the pliability of abstracting over charting grammars by implementing-as a template-the functionality of the shelf builder Polestar (a simulacra of Tableau) and a set of templates that emulate the Google Sheets chart chooser. We show how templates support multimodal visualization editing by implementing a prototype and evaluating it through an approachability study. CCS CONCEPTS• Human-centered computing → Visualization systems and tools; Graphical user interfaces.

show abstract

Section: Validation In Visualmentioning

confidence: 99%

Integrated Visualization Editing via Parameterized Declarative Templates

McNutt

Chugh

2021

Proceedings of the 2021 CHI Conference on Human Factors in Computing Systems

View full text Add to dashboard Cite

show abstract

“…Visual Analytics Process Keim et al 19 Sacha et al 20 Ribarsky and Fisher 22 Federico, Wagner et al 23 Lu et al 15 Lu et al 24 Sacha et al 7 Visualization during preprocessing Heer et al 28 Kandel et al 3 von Zernichow and Roman 29 Kandel et al 16 Visualization of data quality issues Templ et al 32 Eaton et al 33 Sjöbergh and Tanaka 34 Song and Szafir 35 McNutt et al 36 What the practitioners say Milani et al 8 Frequency (... of 17) 8 3 1 0 3 5 9…”

Section: Review and Comparisonmentioning

confidence: 99%

“…Complementing the previous, Item 4 ( Preprocessing impacts in the next phases ) considers the effects that the decisions made during the preprocessing may cause in later stages, similar to the discussion promoted by Crone et al 39 Even though the related work selected as part of Subsections Visualization during preprocessing and Visualization of data quality issues recognize the importance of preprocessing and its impacts on the overall process, most of them are concerned about how to enhance the capabilities of the data analysts while performing the cleaning and transformation tasks. Therefore, only Sacha et al, 7 McNutt et al, 36 and Milani et al 8 mention this topic, at least in an explicit manner. To illustrate, Sacha et al 7 present examples of pathways in the Machine Learning workflow, and during the Evaluate-Model process, they explain that a model-developer may wish to make some changes to what was set in the previous steps, which includes data preparation tasks.…”

Section: Related Workmentioning

confidence: 99%

PrAVA: Preprocessing profiling approach for visual analytics

Milani

Loges

Paulovich

et al. 2021

Information Visualization

View full text Add to dashboard Cite

To accommodate the demands of a data-driven society, we have expanded our ability to collect and store data, develop sophisticated algorithms, and generate elaborated visual representations of the data analysis process outcomes. However, data preprocessing, as the activity of transforming the raw data into an appropriate format for subsequent analysis, is still a challenging part of this process. Although we can find studies that address the use of visualization techniques to support the activities in the scope of preprocessing, the current Visual Analytics processes do not consider preprocessing an equally important phase in their processes. Hence, with this paper, we aim to contribute to the discussion of how we can incorporate the preprocessing as a prominent phase in the Visual Analytics process and promote better alternatives to assist the data analysts during the preprocessing activities. To achieve that, we are introducing the Preprocessing Profiling Approach for Visual Analytics (PrAVA), a conceptual Visual Analytics process that includes Preprocessing Profiling as a new phase. It also contemplates a set of guidelines to be considered by new solutions adopting PrAVA. Moreover, we analyze its applicability through use case scenarios that show resourceful methods for data understanding and evaluation of the preprocessing impacts. As a final contribution, we indicate a list of research opportunities in the scope of preprocessing combined with visualization and Visual Analytics to stimulate a shift to visual preprocessing.

show abstract

“…For instance, Correll et al [20] use simulation of data quality issues to highlight visual designs that may not robustly or reliably surface important properties in distributions. More generally, McNutt et al [49] propose the use of simulation results to automatically detect potentially misleading or unstable insights from visualization. An insight that is highly sensitive to particular conditions may not be reliable or generalizable.…”

Section: Sensitivity Analyses Of Visual Analyticsmentioning

confidence: 99%

“…The former is limited in the amount of detailed changes that can be noticed (an analyst may not notice small changes in proportions of documents belonging to particular topic clusters), and the latter is limited by the amount of the complexity that can be shown at once: term-topic matrices, for instance, often only show a small number of "top" tokens or topics, as it is infeasible to provide information about tens of thousands of tokens in detail in one view, and rely on interactivity or different ordering metrics to opportunistically surface different parts of the dataset [2]. These abstractions and summarizations can result in the potential for AVD "failures" [38] (see Figure 2) or visualization "mirages" [49], where either important updates to the model fail to be represented in a salient way in the resulting visualization, or the visualization of a model may be highly altered visually without much change to the underlying topics or classification accuracy.…”

Section: Visual Analysismentioning

confidence: 99%