Faceting and Small Multiples

When a single chart starts to crowd, the fix is almost never "make the chart bigger." It's make many small charts — one per category, arranged in a grid with shared axes. This pattern is called small multiples (a term coined by Edward Tufte) or faceting.

It is one of the most powerful and underused techniques in modern visualization, and Plotly Express makes it almost free.

The pitch

Small multiples leverage a deep fact about perception: the eye compares parallel pictures faster than it parses one busy picture. Shared axes mean any difference between panels reflects a difference in the data, not the scale.

Faceting by column

facet_col="..." creates one chart per value of that column, arranged horizontally.

You can now compare lunch vs dinner distributions side by side, each split by sex. One chart, two stories.

Faceting by row and column at once

Combine facet_row and facet_col for a full grid:

You now have 4 days × 2 times = 8 mini-scatters, each split by smoker. Eight charts at a glance, with shared axes so every comparison is on the same scale.

Spaghetti chart → small multiples

The classic use of facets is breaking up a too-busy line chart.

facet_col_wrap=3 tells Plotly to wrap to a new row after every 3 panels — handy when you have many categories.

Shared vs independent axes

By default, faceted Plotly Express charts use shared axes: every panel has the same x range and the same y range. This makes comparison straightforward.

Sometimes you want each panel to scale independently — for example, if one country has a tiny range and another has a huge range, sharing axes might flatten the small one.

fig.update_xaxes(matches=None)
fig.update_yaxes(matches=None)

But beware: independent axes mean the reader cannot directly compare values across panels. Each panel becomes a standalone chart. Default to shared axes; use independent only when you have a specific reason.

Why this is so powerful

Small multiples solve three problems at once:

1. Reduce visual clutter. No more 30-line spaghetti.
2. Enable comparison. Shared axes make differences between panels read as differences in the data.
3. Scale visually. A 5×5 grid of small charts shows 25 stories in the same space as one busy chart.

Tufte called small multiples "perhaps the best design solution for a wide range of problems in data presentation." Almost every sophisticated information graphic in major newspapers uses them.

When NOT to facet

• When you have 2 or 3 series — overlap them on one chart.
• When the interaction between groups matters more than the individual groups — overlapping series let you see crossover.
• When the categories aren't comparable — facets only make sense when the panels are "like with like."

Check your understanding