Diff in the Loop: Supporting Data Comparison in Exploratory Data Analysis

Wang, April Yi; Epperson, Will; DeLine, Robert; Drucker, Steven M.

doi:10.1145/3491102.3502123

Cited by 22 publications

(11 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Besides error detection, visualization tools may help analysts interpret analysis pipelines. Recent studies have taken a variety of approaches, such as explaining the analysis pipelines with animations (Datamations [50]), comparing results of alternative data manipulations (DITL [63]) and showing thumbnails of variable distributions inline with code [29].…”

Section: Increasing Transparency and Visibilitymentioning

confidence: 99%

How Data Analysts Use a Visualization Grammar in Practice

Pu¹,

Kay²

2023

Preprint

View full text Add to dashboard Cite

Visualization grammars, often based on the Grammar of Graphics (GoG), have much potential for augmenting data analysis in a programming environment. However, we do not know how analysts conceptualize grammar abstractions, or how a visualization grammar works with data analysis in practice. Therefore, we qualitatively analyzed how experienced analysts (N=6) from TidyTuesday, a social data project, wrangled and visualized data using GoG-based ggplot2 without given tasks in R Markdown. Though participants' analysis and customization needs could mismatch with GoG component design, their analysis processes aligned with the goal of GoG to expedite visualization iteration. We also found a feedback loop and tight coupling between visualization and data transformation code, explaining both participants' productivity and their errors. From these results, we discuss how future visualization grammars can become more practical for analysts and how visualization grammar and analysis tools can better integrate within a programming (i.e., computational notebook) environment.

show abstract

Section: Increasing Transparency and Visibilitymentioning

confidence: 99%

How Data Analysts Use a Visualization Grammar in Practice

Pu¹,

Kay²

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…We propose a concept called Diff in the Loop (DITL) where visualizing differences in datasets is presented as a first-class citizen in data science programming environments [11]. DITL stores a snapshot of the code and runtime variables as users make changes in the code editor.…”

Section: Ditlmentioning

confidence: 99%

“…We have explored various design techniques to help data scientists handle off analysis work during collaboration [9]- [11]. However, as mentioned in the previous study [8], data scientists need better support to improve awareness and avoid conflict editing in real-time collaboration.…”

Section: Future Work: Improving Awareness and Avoid Conflict Editing ...mentioning

confidence: 99%

“…We are interested in exploring context aware mechanisms to support real-time collaborative data science. Same with prior studies [9]- [11], we will take a user-centered approach to understand the design motivations, integrate design goals in the system, and perform a series of user studies to evaluate the performance of the system.…”

Section: Future Work: Improving Awareness and Avoid Conflict Editing ...mentioning

confidence: 99%

“…for generating natural language explanations of the data science code [10]; 3) a concept and prototype of implementing dataframe visualizations as a first-class citizen in data science programming environments to make sense of the impact of code changes [11]. Lastly, I will also present an ongoing project on improving awareness and avoiding conflict editing in real-time collaborative notebooks.…”

mentioning

confidence: 99%

See 2 more Smart Citations

Improving Real-Time Collaborative Data Science Through Context-Aware Mechanisms

Wang

2022

2022 IEEE Symposium on Visual Languages and Human-Centric Computing (VL/HCC)

Self Cite

View full text Add to dashboard Cite

Data scientists benefit from collaboration: data scientists work across disciplines with a variety of stakeholders in practice [1]; data scientists collaborate between the team to improve work efficiency [2]; citizen data scientists collaborate in an open source manner to collectively explore topics of shared interests [3], [4].However, collaboration in data science is often hard. Since data science is highly exploratory [5], the artifact and analysis often iterate fast. It is difficult to maintain a shared understanding across various collaborators. On the other hand, tools like computational notebooks provide a convenient approach for data scientists to run, document, and share analysis in a storytelling way [6]. It satisfies the basic collaboration needs for data scientists to communicate and iterate on each other's work. However, such benefits are rudimentary. There are still many open-ended questions about how to improve the collaboration experience by designing better collaborative data science tools. For example, data scientists often neglect to keep updated documentation during rapid exploration, which results in computational notebooks that are messy and difficult to read [7]; without strategic planning, working together in a shared notebook may block each other's work.My research draws upon human-centered design techniques to identify barriers in real-world data science programming practices, and explore the design space of collaborative data science environments through tool-building. In this paper, I will first review our prior work on understanding how data scientists use computational notebooks for collaboration [8]. Through a mixed-method study, we found that working on the synchronous notebooks improves collaboration by creating a shared context, encouraging more exploration, and reducing communication costs. We also identified several challenges with the synchronous notebook editing tools such as producing messy and less organized notebooks, causing conflict editing without strategic planning.Inspired by the study results as well as related work, we then developed a series of prototypes that aim to help data scientists handle off work during collaboration: 1) a prototype that captures the contextual links between messages and notebook elements [9]; 2) a set of automatic approaches

show abstract

Habitat traits and predation interact to drive abundance and body size patterns in associated fauna

Stelling‐Wood,

Poore,

Hughes

et al. 2023

Ecology and Evolution

View full text Add to dashboard Cite

Habitat‐forming organisms provide three‐dimensional structure that supports abundant and diverse communities. Variation in the morphological traits of habitat formers will therefore likely influence how they facilitate associated communities, either via food and habitat provisioning, or by altering predator–prey interactions. These mechanisms, however, are typically studied in isolation, and thus, we know little of how they interact to affect associated communities. In response to this, we used naturally occurring morphological variability in the alga Sargassum vestitum to create habitat units of distinct morphotypes to test whether variation in the morphological traits (frond size and thallus size) of S. vestitum or the interaction between these traits affects their value as habitat for associated communities in the presence and absence of predation. We found morphological traits did not interact, instead having independent effects on epifauna that were negligible in the absence of predation. However, when predators were present, habitat units with large fronds were found to host significantly lower epifaunal abundances than other morphotypes, suggesting that large frond alga provided low‐value refuge from predators. The presence of predators also influenced the size structure of epifaunal communities from habitat units of differing frond size, suggesting that the refuge value of S. vestitum was also related to epifauna body size. This suggests that habitat formers may chiefly structure associated communities by mediating size‐selective predation, and not through habitat provisioning. Furthermore, these results also highlight that habitat traits cannot be considered in isolation, for their interaction with biotic processes can have significant implications for associated communities.

show abstract

Diff in the Loop: Supporting Data Comparison in Exploratory Data Analysis

Cited by 22 publications

References 30 publications

How Data Analysts Use a Visualization Grammar in Practice

How Data Analysts Use a Visualization Grammar in Practice

Improving Real-Time Collaborative Data Science Through Context-Aware Mechanisms

Habitat traits and predation interact to drive abundance and body size patterns in associated fauna

Contact Info

Product

Resources

About