Sketching is a common practice among visualization designers, and an approachable entry to visualizations for individuals, but moving from a sketch to a full fledged data visualization often requires throwing away the original sketch recreating it from scratch. We aim to instead formalize these sketches, enabling them to support iteration and systematic data mapping through a visual-first templating workflow. In this workflow, authors sketch a representative visualization and structure it into an expressive template for an envisioned or partial dataset, capturing implicit style as well as explicit data mappings. In order to demonstrate and evaluate our proposed workflow, we implement DataGarden, and evaluate it through a reproduction and a freeform study. We discuss how DataGarden supports personal expression, and delve into the variety of visualizations that authors can produce with it, identifying cases which demonstrate the limitations of our approach and discuss avenues for future work.

Video Figure

Evaluation

We evaluated Datagarden using three established methods: a reproduction study with 12 participants to evaluate the system's concepts; a free-form evaluation with 4 visualization experts to evaluate the workflow appropriability and creativity support of the system; and a gallery of visualizations produced with Datagarden to evaluate the strengths and limitations of the data structures and approach.

Gallery

Mood Garden

This visualization demonstrates a multi dimensional mood data set.

✎ Click to open the template in Datagarden

Petri Dish

A simple two category viz. This visualization utilizes a custom spine to achive a scatter layout.

✎ Click to open the template in Datagarden

✎ Click here to open the generated visualization in DataGarden

Blog of Tree

This visualization follows the common stem-branch-leaf pattern of visualization, and demonstrates significant variation in branch shapes.

✎ Click to open the template in Datagarden

Owls

A demonstration of complex glyphs with manually specified spines. Demonstrates the need for overspecification, as position has to be specified to ensure the heads remain on the necks, so to speak.

✎ Click to open the template in Datagarden

✎ Click here to open the generated visualization in DataGarden

Notes

A demonstration of a branch that does not look like a branch. The staffs serve as a branch for the music notes.

✎ Click to open the template in Datagarden

Beads

Experimentation with unusual branch shapes.

✎ Click to open the template in Datagarden

Video Figure

The communication visualization produced in the study figure.

✎ Click to open the template in Datagarden

Dear Data Week 27 adapted to Activity Data

An adaptation of Giorgia Lupi's Dear Data week 27 Media. We simplified the visualization and applied it to activity data to use as the visualization to reproduce in the study.

✎ Click to open the template in Datagarden

Dear Data Week 1 Clocks Recreation

A recreation of Stefanie Posavec's dear data week 1, visualizing checking clocks. The original visualization split the branches into 7 sub branches by day of the week, which was not replicated for simlicity.

✎ Click to open the template in Datagarden

✎ Click here to open the generated visualization in DataGarden

Dear Data Week 3 Lupi Thank Yous Recreation

A recreation of Giorgia Lupi's dear data week 3, visualizing thank yous. This recreation demonstrates the difficulty of reproducing complex layouts.

✎ Click to open the template in Datagarden

✎ Click here to open the generated visualization in DataGarden

Dear Data Week 3 Posavec Thank Yous Recreation

A visualization based off of Stephanie Posavec's dear data week 3, visualizing thank yous.

✎ Click to open the template in Datagarden

Working Ants

This visualization was an experimentation in scaling. We generated a visualization for a full year of work data, with each day represented as a trail (with the length mapped to the length of the work day) and each work activity represented as an ant (positioned based on when the activity finished) of different colors (black are productive activites, grey semiproductive, white unproductive). DataGarden generated and downloaded the visualization image, but the sheer quantity of glyphs rendered the system inoperable.

✎ Click to open the template in Datagarden

Recreation of Expert Participant Designs

Here we reproduce the design by the expert participants. As these designs push the boundries of what the system is capable of, there was insufficient time during the study to complete them, so we reproduced them after the study, and present the recreations here.

E1 Recreation - Music Rating

This visualization displays music rankings, with each song being a bauble, and the position of the bauble displaying the ranking. The color of the bauble indicates it's music genre, and the snowflakes indicate the people who voted for the song.

✎ Click to open the template in Datagarden

E2 Recreation version 1 - Project Tracking

This visualization displays activity data, with each bubble being a project, colored by project category (teaching, research, service, etc). Each smaller bubble is one activity in the project, scaled by the hours it took. The line inside the smaller bubble indicates the completion percentage for the activity.

✎ Click to open the template in Datagarden

E2 Recreation version 2 - Project Tracking

This visualization is a variation on E2's design for activity data, with each bubble being a project, colored by project category (teaching, research, service, etc). Each smaller bubble is one activity in the project, scaled by the hours it took. The progress on the activity is indicated by another circle which fills out the bubble as the activity is completed. As can be seen, laying out this kind of visualization is a challenge.

✎ Click to open the template in Datagarden

✎ Click here to open the generated visualization in DataGarden

E3 Visualization 1 Recreation - Books Read

This visualization displays reading activity, with each shelf being a month, the angle being the completion of the book, and the color linked to how the author rated the book.

✎ Click to open the template in Datagarden

E3 Visualization 2 Recreation - Scatterplot

This visualization is an experiement with producing a scatterplot within DataGarden.

✎ Click to open the template in Datagarden

E4 Recreation - Travel

This shows places visited around the world, with each smaller circle being a country, each dot on the circle being a trip, with color representing the reason, and the glyphs along the trip indicating activities taken part in.

✎ Click to open the template in Datagarden

Study Apparatus

Study 1: Reproduction Study

The reproduction study asked participants to create a visualization given a visual inspiration and set of data mappings. Below are the materials used during the study.

Download Survey The survey provided to participants.
Download Study 1 Protocol Study protocol that was used for conducting the study.
Download Study 1 Task Deck Slide deck that participants were asked to follow.
Download Worksheet The worksheet the participants completed in Task 1.

Resulting Reproductions

Study 2: Free-form Study with Experts

In the freeform study, experts were instructed in the use of the system, then worked through the design of a visualization.

Download Demographic Questions The pre-study demographic interview
Download Tutorial Slides The Slide deck provided to participants to instruct them in the use of the system.
Download Gallery Overview An overview of a few visualization authored with DataGarden to provide inspiration for what is possible with the system.

DataGarden: Formalizing Personal Sketches into Structured Visualization Templates