← All chart types

correlation family

scatter

Two-variable correlation across many points.

Each point encodes two variables (x, y). The cloud shape reveals correlation, clusters, outliers. The fundamental tool for showing relationships between numeric variables.

When to use

30+ observations where the relationship or clusters are the point.

When NOT to use

Few points (use bar or labels), or you only have one variable (use histogram).

Rhetorical roles

  • correlate

Variants & related

Required primitives

  • has_axis_titles

Optional primitives

  • has_legend
  • has_data_labels
  • has_trendline
  • has_source_note

Common anti-patterns

  • hero-with-no-takeaway
  • unsourced-data

Disambiguation rules

  • Loose cloud of dots in 2D space.
  • If a fitted line / curve is drawn through the points, it is scatter-with-trendlines.
  • If point size encodes a third variable, it is bubble.