Proceedings: GI 2017

Raising the Bars: Evaluating Treemaps vs. Wrapped Bars for Dense Visualization of Sorted Numeric Data

Mehmet Adil Yalçın (Keshif, LLC), Niklas Elmqvist (University of Maryland, College Park), Benjamin B. Bederson (University of Maryland, College Park)

Proceedings of Graphics Interface 2017: Edmonton, Alberta, 16-19 May 2017, 41 - 49

DOI 10.20380/GI2017.06

  • Bibtex

    @inproceedings{Yal{\c c}{\i}n:2017:10.20380/GI2017.06,
    author = {Yal{\c c}{\i}n, Mehmet and Elmqvist, Niklas and Bederson, Benjamin},
    title = {Raising the Bars: Evaluating Treemaps vs. Wrapped Bars for Dense Visualization of Sorted Numeric Data},
    booktitle = {Proceedings of Graphics Interface 2017},
    series = {GI 2017},
    year = {2017},
    issn = {0713-5424},
    isbn = {978-0-9947868-2-1},
    location = {Edmonton, Alberta},
    pages = {41 -- 49},
    numpages = {9},
    doi = {10.20380/GI2017.06},
    publisher = {Canadian Human-Computer Communications Society / Soci{\'e}t{\'e} canadienne du dialogue humain-machine},
    keywords = {Visualization; Graphical Perception; Treemaps; Evaluation},
    }

Abstract

A standard (single-column) bar chart can effectively visualize a sorted list of numeric records. However, the chart height limits the number of visible records. To show more records, the bars could be made thinner (which could hinder identifying records individually), and scrolling requires interaction to see the overview. Treemaps have been used in practice in non-hierarchical settings for dense visualization of numeric data. Alternatively, we consider wrapped bars, a multi-column bar chart that uses length instead of area to encode numeric values. We compare treemaps and wrapped bars based on their design characteristics, and graphical perception performance for comparison, ranking, and overview tasks using crowdsourced experiments. Our analysis found that wrapped bars perceptually outperform treemaps in all three tasks for dense visualization of non-hierarchical, sorted numeric data.