Relationship

Correlation heatmap / correlogram

Encodes pairwise correlation coefficients in a matrix.

Plot
Relationship
Multivariate exploration

What it answers

Which numeric variables are associated?

Shape: Coloured variable-by-variable matrix
What it shows: Positive/negative association and correlation clusters.

Example

The question

Which of our weekly measures move together?

Correlation between five weekly measures

Grid of correlations: pink is positive, cyan negative, stronger colour means stronger. Visits and orders 0.8; price and discount −0.6; discount and orders 0.6; price and orders −0.5.

Illustrative data. It belongs to no company.

When to use it

Use it when

Scanning pairwise relationships across several variables.

Avoid it when

When nonlinear associations or causal structure are the actual question.

How to read it

Read each variable pair and its correlation colour/value.

Note

Specify correlation method and missing-data handling.

The data you need

Layout: Wide observation table
Fields: ObservationID: text; V1...Vp: numeric columns
One row is: One observation with several measured numeric variables. What is the grain?
Rules: Align variables by observation. Decide scaling, missing-data handling and which columns are features.
Optional: Optional Group. PCA, clustering and correlations require calculations before rendering their derived results.
Preparation: Compute a specified correlation matrix after deciding missing-data and weighting rules. Do not correlate identifiers.
Units: Correlation is unitless, usually within[-1,1]; label the method. It is not a percentage.

A wide table: one row per item, with a separate column for each measure or period.

Worked example

Given corr(X, Y)=0.8 and corr(X, Z)=-0.2, the X-Y cell is strongly positive and X-Z weakly negative under the chosen method.

ID	V1	V2	V3
A	10	30	5
B	12	20	8
C	18	25	6

All charts in the library