Data Visualization
Data Visualization and Digital Services
Section titled “Data Visualization and Digital Services”Source: My personal notes from 2021-06-16 session with Ontario Digital Service
Why visualize data? improved analysis (insights, broad), effective communication (sharing, simple to understand)
Example: Visualizing a cholera outbreak that was traced to a geographical water pump with clusters around pump location
Part art, part science
Section titled “Part art, part science”Art: no right answer, subject aesthetics, different approaches to display data Science: statistics, calculations
- Communicating complex ideas: be intentional and select visualizations that encourage insights, use communication (e.g. language, accessible methods) appropriate to audience
Design: Gestalt Principle of Visual Perception
Section titled “Design: Gestalt Principle of Visual Perception”- Gestalt = shape/form in German
- Proximity (how close objects in a visual are to each other and their
relationship)
- Examples: Table borders, spacing
- Similarity
- Closure
- Symmetry
- Connection
- Continuity
Declutter
Section titled “Declutter”Remove amount of space taken up on paper/screen:
- Consistent fonts, colours, formatting, add story text
- Remove gridlines
- Simple language
- Readable labels and directly label rather than using legends
- Use whitespace
When to Use Which Charts
Section titled “When to Use Which Charts”- Comparison: Column, Bar, Line, Slope, Dumbbell, Pie
- Composition of a whole: Stacked, Column/bar, Pie, Treemap
- Distribution: Histogram, Density Plot, Scatterplot, line
- Trends: Line, Column, Area
- Relationship: Scatterplot, Bubble
Note: Pie charts are not recommended as people find difficulty in seeing differences in slices, especially with more than 4 slices.
Example:
Accessibility
Section titled “Accessibility”- Tell the truth about data: Visualization can be presented to change truth about data. For example editing axis, scales
- Visualizations must meet accessibility standards (distinguishable colours, font and sizes, alternative text, tables left to right and top to bottom, text box order, (tools) test for accessibility)
Design Issues
Section titled “Design Issues”Q: Related to “Understand Your Audience” - are there any practices you have learned to keep in mind when visualizing data for both English and French / multilingual audiences?
- Send for translation
- French text will take 1.25-1.5 times more space, sometimes design for French first to account for spacing
- Choose simple word selection in other language or provide explanations
Q: How to identify outliers?
Histogram, box plat, or density plot of normal distribution, edges will show outliers.
Q: How to assess default visuals in programs?
Check the defaults meet the principles explained above