-
Notifications
You must be signed in to change notification settings - Fork 0
The Storythings Dataviz Process v1.0
In short: Start with a strong focus, do as much research as you can, organise, summarise and then deliver conclusions in a structured and visually appealing form.
Talk to the series editor and find out: What story do you want to tell? What are the key points to be made? How will the chart be useful for readers? What will they be able to do with it?
Look for open datasets. If none are available, see if there is data you could ask for, or even collect yourself. Take careful notes on where you got information from - it's easy to end up with a mess of numbers and files without clear origins.
Spend some time getting to know the data. Do some exploratory data analysis. Poke around and look for patterns and holes. Write or sketch multiple ideas and storyboards for what a visualisation might look like.
What kind of data do you have? (categorical, hierarchical, relational, temporal, spatial) What relationships are you trying to show? How much interactivity is necessary vs nice to have? Look critically at how others have visualised similar information elsewhere. Bear in mind the goals set in step one. Do not use a pie chart.
Go back and show your work so far to the series editor and creative director to get feedback. Then, flesh out details on your sketches, finish off storyboards and hunt down any extra data needed. Run your designs through Ann K Emery's checklist.
Make decisions about typefaces, colour palettes and other style variables. Check palettes for colour blind readability. Clear any chartjunk. Every pixel should mean something. This step goes after the substance and structure are decided upon.
Feed your data into an appropriate tool and create something. Put it in the article, and consider where else it could be shared (eg Twitter, Imgur, Reddit). Make sure all data used is carefully credited. Consider making the data (and any analysis code) publicly available through Google Sheets or Github.