User:M.G.Nagi

Jarvis-Patrick Clustering

Jarvis-Patrick Clustering is a non hierarchic clustering method based on theShared Nearest Neighbors (SNN), with the goal is to find clusters of different types, shapes and densities in high dimensional data Method: Step 1: SNN specification

Construct an SNN Graph from data matrix as follows: if p and q have each others in the KNN list then create a link between them Step 2: Weighting

weight the links with sim(p,q)=∣∣NN(p) ∪ NN(q)∣∣

where NN(p) and NN(q) are k neighbors of p and q resp. Step 3: Filtering

then filter the edges: remove all edges with weight less than some threshold

Step 4: Clusters

let all connected components be clusters