One of my previous post on XKCD python library was inspired by one of the articles in PLoS Computational Biology.
I do agree with most of their comments plus I put some that I think are also important. Some of the points seem to be common sense but it is important to make them as a checklist that nothing would be missed when producing your great figure.
A checklist for preparing figures:
- Who will be looking at your figure:
- Experts or students…
- How you will present your graphs
- Screen display, electronic form, paper?
- Avoiding complicated graphs for presentations
- Font size appropriate for circumstance
- What is your message?
- One chart type will be better than other
- Think and discuss your plots
- Putting adequate legend and caption
- One should provide much effort to make a figure with a caption to stand on its own.
- One should be able to draw appropriate conclusions based on a figure.
- Optimize plot graphics:
- Check if a scale is adequate on each of the axes
- Not always defaults provide the best solution
- Select colors that would convey a message:
- Care for colorblind is a virtue
- Not all colors are well displayed by projectors
- Show exactly your data:
- Again look at the scales
- Avoid 3D if your data is 2D
- Avoid non-necessary elements
- Better clear message than amazing graphics
- Use tools that suit your needs
- Try to not overcomplicate the graph content