Cluster Profile Radar Chart Calculator

Results

Load sample data or enter a dataset, then select Calculate chart.

Interactive cluster radar chart

Cluster similarity matrix

Interpretation and warnings

Cluster profile summary

Feature statistics

Generated report

1. Dataset input

Paste CSV or TSV data

The first row may contain headers. Cluster labels can be text or numbers.

Upload CSV or TSV

Example dataset

Delimiter

Missing-value handling

Duplicate rows

First row contains headers

Trim spaces from labels

2. Column and feature selection

Cluster label column

Optional row weight column

Profile aggregation

Numeric features

Use Ctrl or Command to select multiple features.

Maximum radar axes

Feature selection rule

Low-variance threshold

Custom feature order

Enter comma-separated feature names. Unlisted features remain afterward.

Feature weights

Weights affect profile distance and difference ranking.

3. Scaling and comparison

Feature scaling

Custom radar minimum

Custom radar maximum

Similarity metric

Reference cluster

Value display

Clamp min-max values from 0 to 1

Use shared radial range

Reverse-scaled features

Useful when lower original values represent stronger performance.

4. Chart appearance

Chart title

Chart subtitle

Chart layout

Profile style

Line shape

Colour palette

Line width

Marker size

Fill opacity

0.18

Legend position

Decimal precision

Show markers

Show values beside points

Show radial grid

Show radial tick labels

Animate chart updates

5. Cluster selection

Inspect a dataset to list clusters.

Formula used

Cluster mean: μ_c,f = Σx_i,f / n_c.

Weighted mean: μ_c,f = Σ(w_ix_i,f) / Σw_i.

Min-max scaling: z = (x − min) / (max − min).

Z-score scaling: z = (x − μ) / σ.

Robust scaling: z = (x − median) / IQR.

Euclidean distance: d(a,b) = √Σw_f(a_f − b_f)².

Cosine similarity: s(a,b) = (a · b) / (||a|| ||b||).

How to use this calculator

Paste a dataset or load an example.
Inspect columns and select the cluster label.
Choose numeric features and an aggregation method.
Choose scaling, comparison, and chart options.
Select clusters and calculate the radar chart.
Review summaries, warnings, differences, and similarities.
Export the chart, table, JSON, or printable report.

Example data format

Cluster	Income	Spending Score	Visits	Support Calls
Premium	92	84	18	2
Value	55	72	12	4
At Risk	46	31	5	9

Frequently asked questions

What does a cluster radar chart show?

It compares several cluster profiles across shared numeric features. Each spoke represents one selected feature. Similar shapes indicate similar feature patterns.

Should features be normalized first?

Normalization is usually recommended when features use different units. It prevents large scales from dominating the chart. Keep original values available for interpretation.

How many features should I display?

Five to twelve features are usually readable. Too many axes create visual clutter. Use the differentiating-feature option when necessary.

Which aggregation method should I choose?

Use means for balanced numeric distributions. Medians reduce sensitivity to extreme observations. Weighted means support unequal observation importance.

What is a dominant cluster feature?

It is a feature where one cluster strongly differs. Dominance depends on scaling and comparison choices. Always inspect original values before conclusions.

How is similarity calculated?

Cosine compares profile direction across features. Euclidean compares straight-line distance between profiles. Manhattan sums absolute feature differences.

Can negative values appear?

Yes, z-score and robust scaling can produce negatives. Plotly radar charts support negative radial values. Custom ranges can improve their readability.

Why are some features excluded?

Non-numeric or low-variance columns may be excluded. Axis limits can also reduce displayed features. Review the warnings for exact reasons.

Can this replace formal cluster validation?

No, radar charts provide descriptive profile comparisons. Use silhouette or Davies–Bouldin scores for validation. Combine visual and statistical evidence carefully.