Machine learning--Clustering series--dbscan algorithm

Dbscan algorithm

　　Basic Concept :(density-based Spatial clustering of applications with Noise)

　　　　Core Object : If the density of a point reaches the threshold value set by the algorithm, it is the core point. (That is, the number of points within the R neighborhood is not less than minpts)
　　　　distance threshold for ε neighborhood : Set radius r
　　　　direct density can reach: If a point P in the R neighborhood of Point Q, and Q is the core point p-q direct density can be reached.
　　　　density: If there is a sequence of points q0, Q1 、... qk, the direct density of any qi-qi-1 is reached, it is said from q0 to qk density can be reached, which is actually a direct density of "transmission." Just like MLM, the development of the downline.

　　　　Density connection : If from a core point p, point Q and point k are density can be reached, then the point Q and point K is density connected.
　　　　Boundary Point : a non-core point belonging to a class that cannot be developed offline
　　　　Noise point : A point that does not belong to any one of the clusters, from any core point is the density is not reached, also known as outliers .

　　

Work flow

Given:

Parameter d: input data set
Parameter ε: Specify RADIUS
Minpts: Density threshold (e.g. 5)

　　　　

Parameter selection:
Radius ε, can be set according to K distance: find the mutation point
K Distance: The given DataSet p={p (i); i=0,1,... n}, calculates the distance between points P (i) and the subset S of Set D, the distance is sorted from small to large, and D (k) is called K-distance.
minpts::k-distance from the value of K, generally take smaller, multiple attempts

Advantage:

• No need to specify the number of clusters
• You can find clusters of any shape
• Good at finding outliers (detection tasks)
• Two parameters is enough.

Disadvantage:

• High-dimensional data can be difficult to do (dimensionality reduction)
• Parameters are difficult to select (parameters have a very large effect on the result)
• Sklearn is slow (data reduction strategy)

　　　　　　

　　　　　　

Machine learning--Clustering series--dbscan algorithm