Model Training & Adaptation
DBSCAN
Density-based spatial clustering of applications with noise
A method that groups closely packed examples together and can leave isolated examples outside the groups.
Example
An AI system finds dense geographic clusters while leaving scattered observations ungrouped.
Why people use it
It finds dense groups without forcing isolated cases into a group.
What you'll hear
“Let isolated points remain outside the clusters.”
What this means for you
Check sensitivity to distance, density settings and feature scaling.
Can you control it?
Developer-only
The people building or running the AI choose this setup. An everyday user generally needs their help to change how this part works.
Common questions
- Can two nearby groups be accidentally joined?
- Yes. A chain of closely spaced examples can connect areas people might prefer to treat as separate.
- Must we specify the number of groups first?
- No. The groups emerge from the chosen density and distance settings.
- Can very different group densities be difficult?
- Yes. One set of settings may not suit both tightly packed and widely spread groups.