Analysis Results
Results appear here after calculation.
Selected Point Summary
Choose data and calculate.
Point Classification
| Point | Coordinates | Neighbours | Status | Cluster | Nearest core | Core distance |
|---|---|---|---|---|---|---|
| No analysis yet. | ||||||
Distance and Neighbourhood Table
| Source | Target | Coordinates | Distance | Inside ε | Target status | Cluster |
|---|---|---|---|---|---|---|
| No distance data yet. | ||||||
Neighbourhood and Cluster Plot
k-Distance Plot
Neighbour Distance Distribution
Neighbour Count by Point
Parameter Sensitivity
Parameter Comparison
| ε | MinPts | Metric | Clusters | Core | Border | Noise | Noise % |
|---|---|---|---|---|---|---|---|
| Run comparison to populate this table. | |||||||
Formula Used
Nε(p) = { q ∈ D | d(p,q) ≤ ε }
A point is core when its epsilon neighbourhood contains at least MinPts points.
A border point is not core but lies inside a core point neighbourhood.
A noise point is neither core nor border.
How to Use
- Enter points manually, paste rows, import CSV, or load an example.
- Choose epsilon, MinPts, distance metric, scaling, and feature weights.
- Select a reference point and calculate all DBSCAN neighbourhoods.
- Inspect classifications, distance tables, clusters, charts, and warnings.
- Compare parameter combinations and export reports or CSV files.
Example Data
| Point | X | Y | Expected role |
|---|---|---|---|
| A1 | 1.0 | 1.2 | Likely core |
| A2 | 1.4 | 1.1 | Likely core or border |
| B1 | 5.2 | 5.0 | Likely core |
| N1 | 8.2 | 1.1 | Likely noise |
Frequently Asked Questions
What does epsilon control?
Epsilon sets the maximum distance for neighbourhood membership.
What does MinPts control?
MinPts defines how many nearby points create a core point.
Should the point itself count?
DBSCAN commonly counts the point itself inside its neighbourhood.
Why do all points become noise?
Epsilon may be too small, or MinPts may be too large.
Why does one huge cluster appear?
Epsilon may be too large for the dataset scale.
Should data be scaled?
Scaling helps when features use very different numeric ranges.
What is a border point?
It is near a core point but lacks enough neighbours itself.
Can DBSCAN find irregular clusters?
Yes. Density connectivity supports curved and non-spherical cluster shapes.
How should epsilon be selected?
Inspect the k-distance curve and choose a meaningful elbow.