dm.cs.tu-dortmund.de/en/mlbits/cluster-density-dbscan-parameters/
DBSCAN Parameterization – Lecture Notes
depending on the application)
Choosing DBSCAN parameters /2
Example: \(\text{minPts}=4\) , therefore plot \(3\) -nearest-neighbor distances:
Sort the distances, look for a knee (subjective!), or choose a small [...] without. On large data, we can query only a subset to make this faster.
Choosing DBSCAN parameters /3
Heuristics on the mouse
nested clusters data set.
only a heuristic — there are no correct parameters [...] DBSCAN revisited, revisited: Why and how you should (still) use DBSCAN . ACM Trans. Database Syst. 42, 3 (2017), 19:1–19:21. DOI: 10.1145/3068335
continue to OPTICS Clustering »
« return to Example for DBSCAN …