3D Scatter Plots

Use 3D carefully

3D plots can look impressive but are often harder to interpret than they seem — a screen is 2D, so any 3D chart is really a projection, and projections distort distance and depth judgment. Two points that look close together might be far apart along the axis pointing “into” the screen.

Use 3D when:

• You genuinely need three continuous numeric variables plotted together
• Interaction (rotate/zoom) helps understanding, and the reader will actually rotate it
• Two 2D scatter plots (or a bubble chart) can’t tell the same story

If you just want to add a third variable to a 2D scatter, prefer color or bubble size (see the previous two pages) — it’s usually clearer than forcing a third spatial axis.

Example

3D Scatter

import pandas as pd
import plotly.express as px
 
df = pd.DataFrame({
    "x": [1, 2, 3, 4, 5],
    "y": [10, 6, 8, 4, 7],
    "z": [100, 120, 90, 150, 110],
    "group": ["A", "A", "B", "B", "A"],
})
 
fig = px.scatter_3d(df, x="x", y="y", z="z", color="group", title="3D scatter")
fig.show()

In the rendered chart, dragging rotates the camera and scrolling zooms — that interactivity is what makes a 3D plot usable at all; a static screenshot of the same figure often hides more than it reveals.

sketch A 3D scatter, projected onto a 2D screen p5.js Pause Restart

Points sit at (x, y, z); depth (z) is faked here with an isometric-style projection and size/brightness falloff.

Tips

• Rotate before you conclude anything — a single static angle can hide clusters or trends.
• Prefer scatter_3d only when a 2D-plus-color/size chart genuinely loses information.
• Heavy point counts get cluttered fast in 3D; consider sampling or aggregating first.

Analyst’s tip

Interactivity does real work here: a 3D scatter plot is nearly useless as a still image, because any single camera angle can accidentally hide a cluster behind another. If you must share a 3D chart as a static export, first rotate to the angle that best shows the pattern — don’t just export the default view.

diagram Should this be a 3D scatter? mermaid

  flowchart TD
  A["Need to show 3 numeric variables"] --> B{"Can color or bubble size
carry the 3rd variable?"}
  B -- "Yes" --> C["Use 2D scatter/bubble chart
(easier to read, no rotation needed)"]
  B -- "No, spatial depth matters" --> D["Use scatter_3d
and expect the reader to rotate it"]

3D adds real value only when a 2D chart with color or size can't capture the pattern.

Next

Continue to: Creating Dashboards with Plotly to combine multiple charts into one view.

🧪 Try It Yourself

Exercise 1 – Build a 3D Scatter

pch.exerciseTag pch.exerciseTitle

Exercise 2 – Color the Third Group

pch.exerciseTag pch.exerciseTitle

Exercise 3 – When to Avoid 3D

pch.exerciseTag pch.exerciseTitle