liveBook · Manning

Reachability

As explained in Data Without Labels

Density reachability is the DBSCAN property by which points are connected through a sequence of core-point neighborhood relationships.

Density reachability connects observations through a sequence of neighborhood relationships in DBSCAN. A point y is directly reachable from x when y lies within ε of the core point x. This direction matters: core points can reach noncore points, but noncore points cannot reach others. section 2.5.2

DBSCAN uses reachability alongside density thresholds:

  • Core points contain at least minPts observations within radius ε, including themselves, and form cluster building blocks. section 2.5.2
  • Border or reachable points are noncore points that can be reached from core points. section 2.5.2
  • Changing ε and minPts changes the density threshold and can alter the resulting connectivity. section 2.5.2

A multi-step connection

If X and Y are core points, a point Z outside X’s immediate neighborhood can still be reached from X through Y. The chain illustrates density-connected points. section 2.5.2

Figure 2.30 X and Y are the core points, and we can travel from X to Y. Though Z is not in the immediate neighborhood of X, we can still reach Z from X through Y. This is the core concept of density-connected points used in DBSCAN clustering.
figure

Trace the path from X to Y and onward to Z to see how DBSCAN connects points beyond a single neighborhood. section 2.5.2

How Data Without Labels covers it

How deep it goes
Explains direct reachability, core-point directionality, and multi-step connectivity, but stops short of a full formal derivation.
Examples
Uses X, Y, and Z to demonstrate a multi-step path beyond one immediate neighborhood.
What it assumes
Assumes familiarity with observations and neighborhoods while defining ε, minPts, core points, and border points.
What you can do after
Understand DBSCAN connectivity by tracing core-point paths and interpreting reachable noncore points.
How it connects
Connects core points, border points, outliers, ε neighborhoods, minPts, and DBSCAN clustering.

Also explained in

Compare how 2 books explain reachability →

This explainer was generated by Manning's knowledge graph from the text of Data Without Labels. The section links point to the parts of the book it draws on.