Tips for good visualization Source: [Link]
ly/2XXDAUv
Know your audience, understand what you’re trying to convey, and how Label things directly wherever possible. Don’t make the reader work
they’ll interact with the information (Online? In a report? On a poster?): to figure out what the orange star means.
• Do you want to explore the data (or let others explore)? Avoid uncommon acronyms wherever you can. If you have room to spell it out, and
• Do you want to foster dialogue between two parties? it’s not something obvious like USAID, do it! This prevents the reader from having
• Do you want to tell a compelling story? to do unnecessary work to search within the text to find what the acronym means.
Simplify, simplify. Eliminate anything you don’t need, like: Be consistent wherever you can:
• Data that aren’t pertinent to your story (Note: this does not mean you get to cherry-pick • Axis limits: if you compare two plots, make sure the x-axis and y-axis have the
which data looks best, but if certain variables are ancillary to the story, don’t include them. same range.
Edit but don’t skew, and be transparent about what you’ve excluded!) • Have colors mean the same thing everywhere.
• Make sure colors have a clear meaning and purpose. Don’t go overboard. • Start axes at zero unless you have a good reason not to.
• Tick marks and axis labels. Use as many as you need to make it clear, but don’t go
overboard. Order your variables in a meaningful way.
• Excess numerical digits (your axis does not need to be labeled $1.894327849 B; $1.9 B will You can also group and transform data. Convert to millions; create a new ratio;
work just fine). calculate a percentage.
• Remove unnecessary grid lines, boxes, and tick marks.
• 3D charts or embellishments: they almost always make things more confusing.
Don’t mislead people with your representation.
• Don’t cherry-pick data. If you exclude data, have a justifiable reason that is more
Fancy software is not needed to make good visualizations (but it can than “It makes the figure look better.”
help). • Cite where you got your information.
• Be clear if info is zero or if it’s missing.
People should be able to understand the whole story by looking at the • Highlight any limitations of the data or uncertainty in their measurement, such as
abstract, figures, and figure captions. All of the following should be good, using error bars.
descriptive, and BIG (it must be readable):
• Titles Sketch and try out different options. Every chart type, color,
• Captions symbol, and placement has its own merits; see what works for
• Annotations your data and your story.
• Axes (with units!)
• Legends (if you have to have one) Most importantly: Good data visualizations will not save you from
• Scale bars (like a temperature gauge)
bad data.
• Sources for your data, if it’s from someone else
Adapted from Geocenter (Laura Hughes). @OpenAcademics