Shape / inspect · All operations

Plot one column against another

A variant of Chart the data

Draw a scatter plot of two numeric columns, or a matrix of scatter plots over several.

Status

In arcform

Not built yet

arcform does not do this yet; it would live in Brightfield's view. Its place on the roadmap

What other tools call it

Meridian's classification; an independent blind rating agreed on 95.7% of 324 rows (variant, version 3).

Alteryx Designer (1)
  • Scatterplotread 2026-09-27

    Plot two fields with optional marginal boxplots, regression line and smooth curve

KNIME Analytics Platform (5)
OpenRefine (1)
  • Scatterplot facetread 2026-09-27

    Plot every pair of numeric columns, linear or log, and filter rows by a rectangle drawn on one plot

VisiData (6)
  • plot-columnread 2026-09-23

    Plot current numeric column against key columns as a terminal scatter graph

  • plot-column-extread 2026-09-23

    Plot current numeric column against key columns with matplotlib/seaborn

  • plot-extread 2026-09-23

    Replot current graph with matplotlib/seaborn

  • plot-numericsread 2026-09-23

    Plot all visible numeric columns against key columns

  • plot-numerics-extread 2026-09-23

    Plot all visible numeric columns with matplotlib/seaborn

  • plot-source-selectedread 2026-09-23

    From Columns Sheet, plot selected source columns against key columns

Background

A scatter plot uses Cartesian coordinates to display values for typically two variables for a set of data.[1] A scatter plot reveals relationships or association between two variables.[2] For k variables, a scatter plot matrix contains all the pairwise scatter plots on a single page, in k rows and k columns.[3]

When to use it

A scatter plot can answer whether two variables are related, linearly or non-linearly, whether the variation in one changes with the other, and whether there are outliers.[2] A scatter plot matrix lets an analyst looking for the strongest relationship scan all the plots and then determine which variables are related.[3] Start from the matrix when you do not yet know which pair of columns matters, then open the single plot of the pair that does. (Meridian's view)

A scatter plot can never prove cause and effect: even though causality implies association, association does not imply causality.[2]

KNIME's Scatter Plot node omits missing values, NaN (not a number) and infinite values, since they cannot be displayed in the view.[4] KNIME's Scatter Plot node has a Max rows option that restricts the number of rows used for the visualisation.[4] Say beside a scatter plot how many rows it left out, whether as missing values or past a row limit, so an empty region is not read as an empty region of the data. (Meridian's view)

The method

In a scatter plot the vertical axis usually holds the response variable, and the horizontal axis a variable suspected of being related to it.[2] If the points of a scatter plot are coded by colour, shape or size, one additional variable can be displayed.[1]

The diagonal of a scatter plot matrix plots each variable against itself, and alternatives put the variable label or a histogram there, or leave it blank.[3] Since each plot of one variable against another mirrors the plot with the axes reversed, some prefer to omit the plots below the diagonal of a scatter plot matrix.[3] In brushing, points selected interactively with a mouse in one plot of a scatter plot matrix are highlighted in the other plots, which are updated dynamically.[3] Brushing is discussed in detail by Becker, Cleveland and Wilks in the paper Dynamic Graphics for Data Analysis.[3]

In OpenRefine, the scatterplot facet previews every number-formatted column plotted against every other, each scatterplot in its own square.[5] In OpenRefine, dragging a rectangle inside a scatterplot facet narrows the data to the rows whose points fall inside the rectangle.[5]

Compared with the alternatives

A conditioning plot, also called a co-plot or subset plot, generates scatter plots of one variable against another dependent on the value of a third variable.[2]

Anscombe's quartet comprises four data sets of eleven (x, y) points that have nearly identical simple descriptive statistics, the correlation between x and y included, yet appear very different when graphed.[6] Anscombe's quartet was constructed in 1973 by the statistician Francis Anscombe to demonstrate the importance of graphing data when analysing it, and the effect of outliers on statistical properties.[6] A correlation coefficient summarises a scatter plot in one number; draw the plot to see what the number leaves out. (Meridian's view)

Sources

  1. [1]Scatter plot - Wikipediasecondary source
  2. [2]1.3.3.26. Scatter Plotsecondary source
  3. [3]1.3.3.26.11. Scatter Plot Matrixsecondary source
  4. [4]Scatter Plotdocumentation
  5. [5]Exploring facets | OpenRefinedocumentation
  6. [6]Anscombe's quartet - Wikipediasecondary source