Best Data Viz
Relationship

Pair Plot / Scatter Matrix

n×n grid of scatter plots for every variable pair, with the diagonal showing each variable's distribution — letting you spot non-linear relationships, clusters, and outliers across all pairs at once before fitting any model.

Car Stats

40 cars across MPG, HP, Weight, Price

View data (40 rows)
Chart data table: Car Stats
MPGHPWeightPrice
32.301332.3620.70
24.602263.2731.40
16.602234.1636.50
29.601022.4416
182514.0847.10
19.701823.6234.70
31.90702.5115
22.601903.4038.10
32.90702.3015
25.602292.6738.40
271832.8433.20
32.30702.0515
282552.1443.70
22.202013.6333
30.401392.7123.60
22.802803.3450.20
21.202713.1852.50
23.802092.7941.60
22.901443.2923.80
29.401422.4826.10
25.401353.3815
36.10701.9015
25.901623.2927
21.401543.5025
231063.2517.10
31.802041.9633.10
30.101972.6029
22.502053.7033
18.801783.8629.90
37.501291.5326.60
23.803323.2455.70
32.70702.3615
23.902102.7738.30
24.901553.6527.60
24.602532.7542.80
23.701593.3228.20
22.302003.2830.40
24703.2915
15.203274.2455.80
19.802023.4028.30
Make a pair plot / scatter matrix with your data

Use a pair plot / scatter matrix when…

  • Exploratory data analysis with 3-8 variables
  • Feature selection
  • Looking for patterns

Avoid when…

  • More than 8 variables (too many panels)
  • Reporting (too complex)

Data it needs

PropertyValue
Min Rows10
Min Columns3
Column Types
numbernumbernumber

Visual anatomy

Marks
circle
Channels
position-xposition-y
Axes
one variable per row, repeated as column

Guiding principles

Consider instead

Common mistakes

  • Too many variables makes the matrix unreadable

  • Inconsistent scales across panels

History

Introduced by Hartigan (1975), popularized by the seaborn Python library.

Accessibility notes

Provide correlation summary table. Describe strongest relationships.

Related reading

Got data? Let's see what works.

Drop your CSV. You'll get a Pair Plot / Scatter Matrix plus four alternatives - ranked by which one actually fits your data best.