Resources Roundup: Making graphs that donāt suck.
A few of my favourite resources, and something to make you stop and think.
So, itās been another week month (I promise Iām working on my publishing frequency!) and Iām back with another resources roundup: this time featuring all the best bits of advice and instruction Iāve seen lately on creating better data visualisations. Enjoy!
Data Visualisation Cheat Sheet (For: Data-Viz Beginners)
The simplest place to start preparing a data presentation is by considering the types of values (and by extension, messages) you want to communicate, and what type of graphs will achieve that best. This DataCamp cheat sheet can help you think about this, down to the subtle differences that even experienced data vizzers might overlook. For example, if you want to represent a trend like daily revenue, you might use a line chart. But if you want to show total sales per day, an area chart might better convey that conceptual difference.
A UX Designerās Guide to Good Data Visualisation (For: everyone)
Choosing the right type of graph is only the first step. In fact, Taras Bakusevych lists that and 19 more ideas for better data visualization, in this UX Collective post which some random guy on the internet called āone of the most useful data visualization guides I have ever seen.ā1 Random guy is right: itās an excellent collection of tips like: Practical ways not to confuse the audience, how to use colours and fonts to highlight your message, how to design for accessibility, and how to do a pie chart properly (if you really, really have to have one)2.
Code Snippets for Common Plots (For: Python beginner-intermediates)
Now that you know what you want to plot, youāll need some code. For anyone new to Python, MatplotLib or Plotly (or who still always has to google for the exact syntax), hereās a nice post featuring code snippets for some of the most common visualisation types: Eight Tips for Effective Data Visualisation, featured on Towards Data Science.
A Guide to Making More Accessible Data Visualisations (For: everyone).
Iām going to add a little more detail to this one than I have/will for the others, as itās important: The Urban Instituteās āDo No Harm Guide: Centering Accessibility in Data Visualizationā, lays out five key principles for making data visualisations accessible for people with differing abilities:
Design with accessibility in mind from the beginning
Accessibility should not be a specialty
There is no established definition for what makes a data visualization accessible
People with disabilities should be involved
There is not a single right answer for writing alt text
To help data practitioners achieve this, the guide is also full of practical advice on topics like designing for cognitive load, writing better alt text, and coding accessible visualisations. If you read just one resource from my post today, I hope itās this.
You can find the guideās editors on twitter: Jonathan Schwabish, Susan J. Popkin, and Alice Feng. When pasting Schwabishās twitter handle into this draft I discovered he also writes The PolicyViz Newsletter, which Iāve bookmarked to dive into. There are also nine different chapter authors you might want to chase up: hereās just one, Frank Elavsky, whose twitter thread introduced me to the guide and neatly summarised his chapter:
Do-s and Donāt-s of Titles, Labels and Explanations (For: everyone)
Speaking of text in graphs, Iām in love with Lisa Charlotte Muthās fantastic piece on how to better use text in data visualisations. She doesnāt just rehash the generic pointers, like āalways include a descriptive titleā; instead, she tells you how to make it informative, engaging, and accessible, and gives tips on how to design it with fonts, sizes, boldness and position, too. Hereās just one example; do go and check it out:
A Final, Controversial (?) Piece of Advice (For: the open-minded)
Confident as I am that the above resource will be universally accepted and praised, Iām willing to take a risk with this final one. Iām going to recommend that maybe - just maybe - you might want to consider adding animation to your data visuals. Hereās an article explaining why. Suffice to say, Iām ā¦ considering it. ;)
Hereās the proof, in case you need it.
Yes, pie charts suck. Until they sometimes donāt.