Skip to content
    Flow-alerta.cab

    DBSCAN

    AI fundamentals & modelsDensity-based Spatial Clustering of Applications with NoiseDensity-based clustering
    Editorially reviewed on: 1 source
    Your reading view

    The short answer · 14 words

    In short

    DBSCAN finds dense groups of any shape and identifies isolated data points as noise.

    Flow-alerta.cab (2026): DBSCAN. https://flow-alerta.cab/en/glossar/dbscan. Editorially reviewed on: 10 October 2026.

    How much detail would you like?

    Practical

    DBSCAN does not require a predefined number of clusters and can detect non-convex shapes. This is suitable for spatial patterns, outlier analysis and customer segments under suitable density assumptions.

    Location data forms several dense supply areas; remote individual points are marked as noise.

    Companies first visualize distances and scaling, test parameter stability and technically test clusters. DBSCAN does not automatically deliver meaningful customer groups.

    At a glance

    Core
    DBSCAN is an unsupervised clustering process. Points with enough neighbors within the radius eps become core points; reachable points can be border points, the rest are considered noise. min_samples specifies the required neighborhood density.
    Use
    DBSCAN does not require a predefined number of clusters and can detect non-convex shapes. This is suitable for spatial patterns, outlier analysis and customer segments under suitable density assumptions.
    Example
    Location data forms several dense supply areas; remote individual points are marked as noise.
    Limits
    Select eps and min_samples unchecked, do not scale features or over-interpret results in high dimensions. Different cluster densities are difficult; the exact counting of min_samples is library-specific.

    In context

    Clustering

    Unsupervised learning method that divides data points into groups based on their similarity or structure, without requiring predefined class labels.

    Questions about this term

    How many clusters do you have to specify?

    Not a fixed number.

    What does noise mean?

    Points that are not assigned to a sufficiently dense region.

    Does it work at all distances?

    The metric must fit professionally and for scaling.

    All details and background

    DBSCAN is an unsupervised clustering process. Points with enough neighbors within the radius eps become core points; reachable points can be border points, the rest are considered noise. min_samples specifies the required neighborhood density.

    ELI5 – Explained simply

    Where many points are close together, the process draws groups; individual far-away points leave it out.

    Why it matters in practice

    DBSCAN does not require a predefined number of clusters and can detect non-convex shapes. This is suitable for spatial patterns, outlier analysis and customer segments under suitable density assumptions.

    Example

    Location data forms several dense supply areas; remote individual points are marked as noise.

    Pitfalls and limits

    • Select eps and min_samples unchecked, do not scale features or over-interpret results in high dimensions. Different cluster densities are difficult; the exact counting of min_samples is library-specific.

    Origin & history

    Ester, Kriegel, Sander and Xu introduced DBSCAN at the KDD conference in 1996 as a density-based algorithm for clusters of any shape and noise.

    Use in mid-sized companies

    • Companies first visualize distances and scaling, test parameter stability and technically test clusters. DBSCAN does not automatically deliver meaningful customer groups.

    Related terms

    • Clustering
    • Outlier detection
    • EPS
    • min_samples
    • Unsupervised learning

    Matching Flow-alerta service

    AI opportunity analysis & roadmapData analysis and AI consulting: develop suitable characteristics, distance measure, parameter search and professional validation of clusters.

    Matching tool / use case

    View use casesscikit-learn DBSCAN; Use Case: group spatial service cases and separately examine isolated cases.

    Sources

    • aaai.orgEster et al., „A Density-Based Algorithm for Discovering Clusters in Large Spatial Databases with Noise“, KDD 1996 (Abruf 06.10.2026)

    Quellenprüfung: 10.10.2026. Definition und dokumentierte Einordnung abgeglichen; Beispiele beschreiben mögliche Anwendungen.

    Editorially reviewed on:

    Next step

    Put knowledge into practice

    Discuss your projectView use cases