Imposing connectivity constraints adds additional requirements in regards to which observations can be connected and merged into an existing cluster beyond just the distance metric used. Without connectivity constraints, clusters are merged simply based on satisfying the linkage criteria chosen. In high-dimensional data, not using connectivity constraints can fail to distinguish between regions that are similar based on distance but might practically represent different constructs within the data. A simple example of a connectivity constraint is to restrict candidate observations that can be merged into a cluster to a pre-defined number of observations or nearest neighbors.
How does imposing connectivity constraints help with Agglomerative clustering?
Help us improve this post by suggesting in comments below:
– modifications to the text, and infographics
– video resources that offer clear explanations for this question
– code snippets and case studies relevant to this concept
– online blogs, and research publications that are a “must read” on this topic
Partner Ad
