Dbscan Clustering algorithm based on density clustering algorithm

A: Basic concepts

1.: Object o is centered with O, space for radius, parameter, is the domain radius value that the user specifies for each object.

2.MinPts (field density threshold): The object's number of objects.

3. Core object: If the object number of object o contains at least minpts objects, then the object is a core object.

4. Direct density up to: If the object P is within the core object Q, then p is the direct density from Q can be reached.

5. Density up to: in Dbscan, P is from Q (Core object) density can be reached, if there is an object chain, making, is from about and minpts direct density can be reached, that is, in the

, the density is up to.

6. Density connection: If the object is present, so that the object is from Q about and minpts density can be reached, it is called about and minpts density is connected.

Two: density up to and density connected

Figure 1 Density-to-density connectivity in density-based clustering

The above figure shows that M,P,O.R are core objects because they all contain only 3 objects.

1. The object q is accessible from M direct density. The object m from the p direct density can be reached.

2. The object q is from the P (indirect) density can be reached, because Q from M direct density can reach, m from P direct density can be reached.

3.R and S are available from o density, while O is from R density, all o,r and s are density-linked.

Flow of the Dbscan algorithm: