Skip to main content

Table 1 DRAGON Method

From: Divisive hierarchical maximum likelihood clustering

 1. Given a sample set χ s (at the beginning χ s  = χ), compute likelihood L (Eq. (12)).
 2. Until L * > L, remove one sample \( \widehat{\mathbf{x}}\in {\chi}_s \) (Eq. (18)), compute new likelihood L *, update χ s and L.
 3. Find centroid μ s  = E[χ s ] and d x  = δ(x, μ s ) xχ.
 4. Partition {d x } into two groups, for example using k-means algorithm (or divide into two groups based on their values). One of these groups will have lower d x values (representing closeness to μ s ) whereas the other will have higher d x values (representing distance from μ s ). Update χ s by replacing it with the samples with the lower d x values.
 5. If required repeat steps 3 and 4. Take out the cluster χ s from χ. Update χ accordingly (the updated χ would contain all the samples except χ s ; i.e. χ ∩ χ s  = Φ).
 6. Repeat all the steps until all the possible clusters (or desired number of clusters) are obtained.