dm.cs.tu-dortmund.de/mlbits/cluster-various/
Further Clustering Approaches – Lecture Notes
y)}{n_i^{1+2f\! (\theta )}}\]
merge clusters with best goodness
\[g(C_i, C_j)=\tfrac { \sum \nolimits _{x\in C_i} \sum \nolimits _{y\in C_j}\mathrm{link}(x,y) }{ (n_i+n_j)_{\phantom{i}}^{1+2f\! (\theta [...] (\theta )}-n_i^{1+2f\! (\theta )}-n_j^{1+2f\! (\theta )} }\]
\(f\! (\theta )\) is application and data dependent example in market basket analysis: \(f\! (\theta )=\tfrac {1-\theta }{1+\theta }\)
CHAMELEON [ [...] is decided indirectly, by the diagonal values \(s(i,i)\) .
runtime complexity: \(O(n^2)\) , and needs \(3\cdot n^2\) memory to store \((s_{ij})\) , \((r_{ij})\) , \((a_{ij})\) .
Other Density-based Clustering …
