DVSS - Day 1

Day 1 on the fly notes of MaLGa's Data Visualization Summer School 2025.

Scientific communication

• science != scientific literacy (content/research vs process/impact/methods)
• assess/know the scientific literacy level of the audience

Giorgia Lupi

https://www.informationisbeautifulawards.com/showcase/204-nobels-no-degrees

• peculiar use of space: chart diagonal orientation vs text lexicographic order, to convey information (e.g. "this should be read" vs "this should be seen") and to grab attention
• sample chart as legend
• Approximating the components of Lupi's Nobel, no degrees

Federica Fragapane

https://www.behance.net/gallery/70033395/The-Most-Violent-Cities/

• lollipop chart with polar plot at each marker to convey both relative and absolute magnitude at the same time
• design choices to empathize with data meaning (e.g. encode murder victims as shell lines instead of radius amplitude)

betterposter

https://mitcommlab.mit.edu/be/2023/09/27/toward-an-evenbetterposter-improving-the-betterposter-template/

Our World In Data

https://ourworldindata.org/brief-history-of-ai

Data Visualization

• Anscombe's quartet

John Snow

https://storymaps.arcgis.com/stories/59a6e61a0a61448699f67e29bd45714c

Ed Hawkins

https://showyourstripes.info/

Guidelines for data viz

from "Better Data Visualization"

1. Show the data

• sometimes is enough (e.g. spatial data)

2. Reduce the clutter

• avoid useless visual elements (i.e. which don't convey additional information)

https://datavizproject.com/data-type/bar-chart/

3. Integrate graphics and text

• legend can be placed close to data traces (ref. proximity)
• title, subtitle, text annotations
• choose an appropriate colormap (e.g. default jet colormap in Windy for temperature, which is not a gradient of anything)
• The misuse of colour in science communication

ADD_CAPTION_HERE

• avoid inconsistencies (e.g. legend data sorted differently from data traces)
• add explainers (e.g. clearly state what the takeaway should be)
• chart-in-chart to explain details (zoom-in effect)
• explain how to read the graph

https://nightingaledvs.com/connected-scatterplots-make-me-feel-dumb/

4. Avoid spaghetti charts

• multivariate data can be visualized as a collection of univariate charts (e.g. linecharts in a grid, with fixed axes ranges)

5. Start with gray

• start with everything in the background, and let emerge the important elements only

Visual perception and design principles

Gestalt principles

1. Proximity

• closeness

2. Similarity

• shapes

3. Closure

• tendency to perceive elements even if they are not visible

4. Common fate

• movement

5. Continuity

6. Good figure (Prägnanz)

• visual perception privilege simple and "good" shapes

7. Past experience

Encoding information

https://www.datavizhandbook.info/

Color

• Chroma.js Color Palette Helper

Typography

• The Equilateral Triangle of a Perfect Paragraph

Resources

• A Comprehensive Guide to Unlocking Your Data's Potential
• Royal Statistical Society Best Practices for Data Visualisation