A visualization support tool for advanced hierarchical clustering analysis. MLCut allows cutting dendrograms at multiple heights/levels. In other words, it allows to set multiple local similarity thresholds in potentially large dendrograms. It uses two coordinated views, one for the dentrogram (radial layout), and another for the original multidimensional data (parallel coordinates). The purpose is to add flexibility and enforce transparency in the process of selecting branches that correspond to the different clusters, while enabling the discovery of visual patterns in the original data.
https://research-repository.st-andrews.ac.uk/handle/10023/9518
