Part of a whole

Dendrogram

Shows how observations or groups merge under hierarchical clustering.

Diagram
Hierarchy
Clustering

What it answers

Which observations merge and at what dissimilarity?

Shape: Branches with merge heights
What it shows: Nested similarity groups and candidate cuts.

Example

The question

Which stores are most alike on their sales mix?

Stores clustered by similarity of sales mix (dendrogram)

Stores join where their branches meet; lower joins mean more alike. Malmö and Lund join first (1.2), then Ystad and Kalmar (1.8), Växjö and Visby (2.4). Cutting at 3 gives three groups.

Illustrative data. It belongs to no company.

When to use it

Use it when

Hierarchical clustering with defined distance and scaling.

Avoid it when

Treating merge height as probability or omitting preprocessing.

How to read it

Read leaves, then compare merge heights under the chosen linkage.

Note

Specify distance, linkage and leaf-order conventions.

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: Select/scale features, calculate distances and hierarchical linkage, then retain merge heights and the chosen leaf ordering.
Units: Merge height uses the selected dissimilarity/linkage scale, not probability or necessarily original feature units.

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

Worked example

Given distance(A, B)=1 and distances to C=5, single-linkage mergesA-B at 1 and then joinsC at 5.

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

All charts in the library