[reference] · ⏳ outline — deep note in progress (part of the research-course build).
Readers look at your figures before they read your words — often a figure and the abstract are all they read. A great figure can carry the whole contribution; a bad one can sink a strong paper. This is how to make visuals that argue for you, honestly.
Will cover
- The “figure-first” reality — design the key result as a figure a skimmer can understand in 10 seconds; the caption should stand alone.
- Figures vs tables — figures for trends/comparisons/relationships, tables for exact numbers and many-way comparisons; don’t make readers do the trend-spotting a chart would do for them.
- The anatomy of a good scientific figure — clear axes + units, legible fonts, a self-contained caption, error bars/variance shown, no chartjunk.
- Honest visualization — truncated axes, misleading scales, cherry-picked ranges — the ways charts lie, and why it’s an integrity issue, not just aesthetics.
- Comparison discipline — same axes/scale across compared plots, consistent colors for the same method throughout the paper.
- Accessibility — colorblind-safe palettes, readable in grayscale (many print your paper B&W), high-DPI/vector (PDF/SVG) not blurry PNGs.
- Tooling — matplotlib/seaborn, ggplot, TikZ/PGFPlots for LaTeX-native figures; the
datavizprinciples apply.
Related
Jump to a subject
Software engineering
Infrastructure & security
Computer science
Hardware & engineering