Affordances, originating in psychology, describe how an object's design influences the physical and cognitive actions users may take. Past work applied affordance theory to visualization to explain how design decisions can impact the cognitive actions of visualization readers. In this work, we demonstrate that affordances can complement effectiveness rankings by further explaining the root causes behind visualizations' task performance. To do so, we conduct a case study on static normal probability density function plots, identifying their current affordances. Next, we identify the optimal affordances for a common probability-comparison task and develop a novel affordance-driven visualization, the Croissant Chart, to support them. We empirically validate the design's effectiveness through a preregistered study (n = 808), demonstrating how affordances can inform predictable changes in task performance. Our findings underscore the potential for affordance-based approaches to enhance visualization effectiveness and inform future design decisions.

Objectives Design and test an alternative visualization of Patient-Reported Outcomes Measurement Information System (PROMIS) scores to improve visualization efficacy by addressing common confusion surrounding the direction of improvement when reading line charts of PROMIS symptom scores over time. Materials and Methods We designed a “smiley”-timeline visualization that encodes scores via color and pictographs instead of y-axis direction. We compared trend interpretation when using smiley-timelines vs. line charts via a preregistered, online, randomized, between-subjects experiment with pair-wise comparisons of charts. Participants were recruited via Prolific.com, fluent in English, over age 45, and able to interpret stimuli correctly without initial instruction, allowing for analysis of differences in response times without explicit instruction or learning time confounding results. Results We analyzed 383 participant responses. Color-encoded line charts in which “up” indicated improvement in symptoms and function resulted in the quickest response times. Smiley-timeline response times were second-fastest and not significantly slower. Color-encoded line charts resulted in faster response times than their grayscale counterparts. Smiley-timelines ranked significantly higher than line charts in ease of use, intuitiveness, perceived speed to use, and likelihood of being recommended. Conclusion Color encoding and pictographic timelines may address confusion stemming from longitudinal symptom score line charts. We provide evidence that smiley-timelines are more intuitive than line charts and may be a viable depiction of PROMIS scores. Additional user testing is necessary to determine performance when increasing the number of scores per chart and the number of charts shown. Impact Statement Our research advances the viability of PRO score use by addressing a pervasive problem with interpretability of PRO symptom scores by assessing the comparative effectiveness of an alternate visualization method, a new color-encoded smiley-timeline. The results inform concrete design decisions that can make PROMIS information, and potentially other PRO scores, easier for clinicians and patients to understand and use.

Tens of thousands of people have represented data by creating data-encoding textile pieces like blankets, scarves, and more. A prototypical example is the temperature blanket, which represents the weather through rows or blocks of different colors mapped to temperature ranges. While researchers have used fiber arts mediums to create exploratory projects, data visualization and physicalization research has largely not engaged with examples from this enormous and diverse community. We explore the space of data textiles, or fiber arts that encode information, by surveying creators (i.e., data fiber artists) on their projects and processes. We create a corpus of 159 examples of data textiles and present a schema characterizing the data encoding methods used in these projects. We also gather insights into creators' data workflows as well as their motives and discoveries through making with their data. Creators of data textiles use distinct processes to map their data, building fabric from component structures and substructures while using material properties like color and texture. From diverse data-tracking procedures, creators use and relate to data in varied ways. Working on these pieces also contributes to the creators' personal growth and data understanding. Our findings point to new opportunities for visualization, including opportunities to support fiber artists with tools formatted to their needs and opportunities to incorporate concepts from data textiles into other types of visualization (e.g., using texture, structural layouts, colorways).

What do we value in our visualizations, and in the people who design them? Despite a growing body of work on critical data visualization, the conception of what it is to do ethical data visualization work can often be narrow (for instance, holding that our ethical duties are discharged merely by avoiding overtly lying or manipulating data), or entangled with potentially problematic implicit value structures (such as the assumption of the objectivity and neutrality of data, and so the designer's role being merely the passive conveying of numbers as efficiently as possible). Yet, what it means to act ethically in data visualization is broad and multifaceted, and the virtues to which we should aspire as data visualization researchers and designers are worth explicating. We conducted an interview study with a broad spectrum of 20 experienced data visualization researchers, practitioners, and data artists to solicit their values and ethical considerations around doing visualization work. We report on a list of 68 values, organized into nine virtue clusters, that we encountered in our interviews. These virtues and values together describe a diverse space of matters of care and concern in data visualization: from the unease around the best use of visualization as a tool for persuasion, to the tightrope that visualization practitioners often walk between their professional responsibilities and their personal moral commitments. The virtues themselves, as well as our interviewees' reflections on ethical practice, offer practitioners, researchers, and educators in data visualization a richer vocabulary for ethical reflection and provide a broader foundation for considering and applying visualization ethics.

Classically, affordance research investigates how the shape of objects communicates actions to potential users. Cognitive affordances, a subset of this research, characterize how the design of objects influences cognitive actions, such as information processing. Within visualization, cognitive affordances inform how graphs' design decisions communicate information to their readers. Although several related concepts exist in visualization, a formal translation of affordance theory to visualization is still lacking. In this paper, we review and translate affordance theory to visualization by formalizing how cognitive affordances operate within a visualization context. We also review common methods and terms, and compare related constructs to cognitive affordances in visualization. Based on a synthesis of research from psychology, human computer interaction, and visualization, we propose a framework of cognitive affordances in visualization that enumerates design decisions and reader characteristics that influence a visualization's hierarchy of communicated information. Finally, we demonstrate how this framework can guide the evaluation and redesign of visualizations.

Partial dependence plots (PDPs) and individual conditional expectation (ICE) plots are visualizations used for explaining the behavior of machine learning (ML) models trained on tabular datasets. They show how the values of a feature or pair of features impact a model’s predictions. However, in models with a large number of features, it is impractical for an ML practitioner to analyze all possible plots. To address this, we present new techniques for ranking and filtering PDP and ICE plots and build upon existing strategies for clustering the lines in ICE plots. Together, these techniques aim to help ML practitioners efficiently explore PDP and ICE plots and identify interesting model behavior. We integrate these techniques into PDPilot, a visual analytics tool that runs in Jupyter notebooks. We use PDPilot to study how 7 ML practitioners utilize the ranking, filtering, and clustering techniques to analyze an ML model.

Clustering is an essential technique across various domains, such as data science, machine learning, and explainable artificial intelligence. Information visualization and visual analytics techniques have been proven to effectively support human involvement in the visual exploration of clustered data to enhance the understanding and refinement of cluster assignments. To support the human involvement, several perceptual studies and visual quality metrics have already been proposed. However, the visual perception of clustering quality metrics, also known as Cluster Validity Indexes (CVIs), still remains to be further explored. This paper presents the first attempt of a deep and exhaustive evaluation of the perceptive aspects of clustering quality metrics, focusing on the Davies-Bouldin Index, Dunn Index, Calinski-Harabasz Index, and Silhouette Score. Our research is centered around two main objectives: a) assessing the human perception of common CVIs in 2D scatterplots and b) exploring the potential of Large Multimodal Models, in particular GPT-4o, to emulate the assessed human perception. To this end, we conducted two systematic data studies and a user study covering a broad collection of datasets. By discussing the obtained results, highlighting limitations, and areas for further exploration, this paper aims to propose a foundation for future research activities.

As wake steering is implemented on large wind plants, methods for evaluating how well these systems optimize plant performance are needed. The data analysis from experiments conducted so far generally focuses on evaluating the improvement in power or energy production of the plant and comparing that to the predicted improvement. In this study, we explore a new approach to provide additional insight and validation of optimization tools by measuring the wake deflection experimentally using only the downstream turbine as a sensor. This approach is demonstrated using the SMARTEOLE wind plant wake steering experimental data. Light gradient boosting machines (LGBM) and Gaussian process (GP) machine learning models are trained and then interrogated by making a set of predictions under defined conditions to determine the wake deflection observed at the downstream turbine. The data set is sparse, particularly for larger yaw angles, and noisy due to factors not captured in the SCADA leading to relatively high uncertainty in the predictions. The GP model is better suited to smoothly fit this data. Using the GP model, a wake deflection of 0.35 rotor diameters at 3.7 rotor diameters downwind is estimated for a yaw error of 20 degrees on the upwind turbine.

Data visualizations present a massive number of potential messages to an observer. One might notice that one group's average is larger than another's, or that a difference in values is smaller than a difference between two others, or any of a combinatorial explosion of other possibilities. The message that a viewer tends to notice – the message that a visualization ‘affords’ – is strongly affected by how values are arranged in a chart, e.g., how the values are colored or positioned. Although understanding the mapping between a chart's arrangement and what viewers tend to notice is critical for creating guidelines and recommendation systems, current empirical work is insufficient to lay out clear rules. We present a set of empirical evaluations of how different messages-including ranking, grouping, and part-to-whole relationships–are afforded by variations in ordering, partitioning, spacing, and coloring of values, within the ubiquitous case study of bar graphs. In doing so, we introduce a quantitative method that is easily scalable, reviewable, and replicable, laying groundwork for further investigation of the effects of arrangement on message affordances across other visualizations and tasks. Pre-registration and all supplemental materials are available at https://osf.io/np3q7 and https://osf.io/bvy95 , respectively.

The use and creation of machine‐learning‐based solutions to solve problems or reduce their computational costs are becoming increasingly widespread in many domains. Deep Learning plays a large part in this growth. However, it has drawbacks such as a lack of explainability and behaving as a black‐box model. During the last few years, Visual Analytics has provided several proposals to cope with these drawbacks, supporting the emerging eXplainable Deep Learning field. This survey aims to (i) systematically report the contributions of Visual Analytics for eXplainable Deep Learning; (ii) spot gaps and challenges; (iii) serve as an anthology of visual analytical solutions ready to be exploited and put into operation by the Deep Learning community (architects, trainers and end users) and (iv) prove the degree of maturity, ease of integration and results for specific domains. The survey concludes by identifying future research challenges and bridging activities that are helpful to strengthen the role of Visual Analytics as effective support for eXplainable Deep Learning and to foster the adoption of Visual Analytics solutions in the eXplainable Deep Learning community. An interactive explorable version of this survey is available online at https://aware‐diag‐sapienza.github.io/VA4XDL .

SliceLens is a tool for exploring labeled, tabular, machine learning datasets. To explore a dataset, the user selects combinations of features in the dataset that they are interested in. The tool splits those features into bins and then visualizes the label distributions for the subsets of data created by the intersections of the bins. SliceLens guides the user in determining which feature combinations to explore. Guidance is based on a user-selected rating metric, which assigns a score to the subsets created by a given combination of features. The purpose of the metrics are to detect interesting patterns in the subsets, such as subsets that have high label purity or an uneven distribution of errors. SliceLens uses the metrics to guide the user towards combinations of features that create potentially interesting subsets in two ways. First, SliceLens assigns a rating to each feature based on the subsets that would be created by selecting that feature. This incremental guidance can help the user determine which feature to select next. Second, SliceLens can suggest combinations of features ranked according to the chosen metric, which the user can then cycle through.

People worldwide use SARS-CoV-2 (COVID-19) visualizations to make life and death decisions about pandemic risks. Understanding how these visualizations influence risk perceptions to improve pandemic communication is crucial. To examine how COVID-19 visualizations influence risk perception, we conducted two experiments online in October and December of 2020 (N = 2549) where we presented participants with 34 visualization techniques (available at the time of publication on the CDC’s website) of the same COVID-19 mortality data. We found that visualizing data using a cumulative scale consistently led to participants believing that they and others were at more risk than before viewing the visualizations. In contrast, visualizing the same data with a weekly incident scale led to variable changes in risk perceptions. Further, uncertainty forecast visualizations also affected risk perceptions, with visualizations showing six or more models increasing risk estimates more than the others tested. Differences between COVID-19 visualizations of the same data produce different risk perceptions, fundamentally changing viewers’ interpretation of information.

Model interpretation is increasingly important for successful model development and deployment. In recent years, many explanation methods are introduced to help humans understand how a machine learning model makes a decision on a specific instance. Recent studies show that contextualizing an individual model decision within a set of relevant examples can improve the model understanding. However, there is a lack of systematic study on what factors are considered when generating and using the context examples to explain model predictions, and how context examples help with model understanding and debugging in practice. In this work, we first identify a taxonomy of context generation and summarization through literature review. We then present Context Sight, a visual analytics system that integrates customized context generation and multiple-level context summarization to assist context exploration and interpretation. We evaluate the usefulness of the system through a detailed use case. This work is an initial step for a set of systematic research on how contextualization can help data scientists and practitioners understand and diagnose model behaviors, based on which we will gain a better understanding of the usage of context.

Model interpretation is increasingly important for successful model development and deployment. In recent years, many explanation methods are introduced to help humans understand how a machine learning model makes a decision on a specific instance. Recent studies show that contextualizing an individual model decision within a set of relevant examples can improve the model understanding. However, there is a lack of systematic study on what factors are considered when generating and using the context examples to explain model predictions, and how context examples help with model understanding and debugging in practice. In this work, we first identify a taxonomy of context generation and summarization through literature review. We then present Context Sight, a visual analytics system that integrates customized context generation and multiple-level context summarization to assist context exploration and interpretation. We evaluate the usefulness of the system through a detailed use case. This work is an initial step for a set of systematic research on how contextualization can help data scientists and practitioners understand and diagnose model behaviors, based on which we will gain a better understanding of the usage of context.

Data visualization is powerful in persuading an audience. However, when it is done poorly or maliciously, a visualization may become misleading or even deceiving. Visualizations give further strength to the dissemination of misinformation on the Internet. The visualization research community has long been aware of visualizations that misinform the audience, mostly associated with the terms "lie" and "deceptive." Still, these discussions have focused only on a handful of cases. To better understand the landscape of misleading visualizations, we open-coded over one thousand real-world visualizations that have been reported as misleading. From these examples, we discovered 74 types of issues and formed a taxonomy of misleading elements in visualizations. We found four directions that the research community can follow to widen the discussion on misleading visualizations: (1) informal fallacies in visualizations, (2) exploiting conventions and data literacy, (3) deceptive tricks in uncommon charts, and (4) understanding the designers' dilemma. This work lays the groundwork for these research directions, especially in understanding, detecting, and preventing them.

One of the potential solutions for model interpretation is to train a surrogate model: a more transparent model that approximates the behavior of the model to be explained. Typically, classification rules or decision trees are used due to their logic-based expressions. However, decision trees can grow too deep, and rule sets can become too large to approximate a complex model. Unlike paths on a decision tree that must share ancestor nodes (conditions), rules are more flexible. However, the unstructured visual representation of rules makes it hard to make inferences across rules. In this paper, we focus on tabular data and present novel algorithmic and interactive solutions to address these issues. First, we present Hierarchical Surrogate Rules (HSR), an algorithm that generates hierarchical rules based on user-defined parameters. We also contribute SuRE, a visual analytics (VA) system that integrates HSR and an interactive surrogate rule visualization, the Feature-Aligned Tree, which depicts rules as trees while aligning features for easier comparison. We evaluate the algorithm in terms of parameter sensitivity, time performance, and comparison with surrogate decision trees and find that it scales reasonably well and overcomes the shortcomings of surrogate decision trees. We evaluate the visualization and the system through a usability study and an observational study with domain experts. Our investigation shows that the participants can use feature-aligned trees to perform non-trivial tasks with very high accuracy. We also discuss many interesting findings, including a rule analysis task characterization, that can be used for visualization design and future research.

The prevalence of inadequate SARS-COV-2 (COVID-19) responses may indicate a lack of trust in forecasts and risk communication. However, no work has empirically tested how multiple forecast visualization choices impact trust and task-based performance. The three studies presented in this paper (N = 1299) examine how visualization choices impact trust in COVID-19 mortality forecasts and how they influence performance in a trend prediction task. These studies focus on line charts populated with real-time COVID-19 data that varied the number and color encoding of the forecasts and the presence of best/worst-case forecasts. The studies reveal that trust in COVID-19 forecast visualizations initially increases with the number of forecasts and then plateaus after 6-9 forecasts. However, participants were most trusting of visualizations that showed less visual information, including a 95% confidence interval, single forecast, and grayscale encoded forecasts. Participants maintained high trust in intervals labeled with 50% and 25% and did not proportionally scale their trust to the indicated interval size. Despite the high trust, the 95% CI condition was the most likely to evoke predictions that did not correspond with the actual COVID-19 trend. Qualitative analysis of participants’ strategies confirmed that many participants trusted both the simplistic visualizations and those with numerous forecasts. This work provides practical guides for how COVID-19 forecast visualizations influence trust, including recommendations for identifying the range where forecasts balance trade-offs between trust and task-based performance.