Mixed Correlation Matrix

Combine Pearson r values, correlation ellipses, variable labels, and significance markers in one matrix. Mixed correlation is a Plus feature; standard Pearson correlation heatmaps remain Free.

Prepare the data

Start with a wide table: each column is a numeric variable and each row is the same sample across variables. Do not submit a precomputed correlation matrix as raw observations.

This practice table illustrates structure only and is not evidence for a statistical conclusion. Use sample IDs to align rows; do not select them as correlation variables.

Practice data · not the screenshot source
Sample ID	Metric A	Metric B	Metric C
S1	10	18	7
S2	12	24	6
S3	15	27	9
S4	18	35	5
S5	20	34	4
S6	23	42	8

Create a mixed correlation matrix

Confirm paired observations before choosing variables, groups, and display settings.

Step 1

Input structure

Choose Mixed Correlation Matrix in the heatmap family menu and use Multiple Variables input.

Settings reference · not an interactive panel

Start here: Multiple Variables
Alternative: XYZ Curve Groups

Step 2

Data relationships

Select two or more numeric columns and check that each row refers to the same sample.

Settings reference · not an interactive panel

One column: One variable
One row: The same sample
Select: At least two numeric variables

Step 3

Preview and groups

Optionally give related variables the same group label. Review valid and unavailable correlation counts, then create.

Settings reference · not an interactive panel

One-level groups: Give related variables the same label
Blank label: No group
Before creation: Review valid and unavailable counts

Step 4

Reading the matrix

Use the display, labels, and color pages to refine ellipses, significance markers, groups, and the colorbar.

Settings reference · not an interactive panel

Lower triangle: Pearson r values
Upper triangle: Direction and shape encode correlation
Diagonal: Variable names
Cross: Unavailable, not a significance result

Interpretation and pre-export checks

Correlation does not imply causation. Check outliers, nonlinearity, and sample size before interpreting Pearson r.
A cross means the correlation is unavailable, not zero or non-significant. Check constant columns and insufficient valid pairs.
Significance markers require available p values. Report the threshold; stars are not a measure of effect size.
Variable groups are display labels, not sample grouping or automatic clustering.
XYZ Curve Groups is a separate input mode: map X, Y, and value columns and choose profiles by X or Y. Do not confuse this with row-wise sample pairing in a wide table.

Read the 2D color map tutorial →

Adjustable settings

These are the common settings for this plot type. For detailed line, marker, axis, legend, and font styling, see Styling.

Matrix Layout: lower values / upper ellipses / diagonal labels
Ellipse size, opacity, and outline
Cell grid
Significance markers and threshold
One-level row and column groups
Palette and Pearson r colorbar

Styling →