Just want to get verified with hierarchical clustering.
- Let's assume that you want dissimilarity measure as euclidean distance.
Let's say that you have three clusters. Each cluster has many observations. If you measure dissimilarity by single linkage, then you are going to fuse the 2 clusters that contain the 2 closest pair of observations/ elements which means these 2 clusters are alike.
If you measure dissimilarity by the complete linkage, you are going to fuse the 2 clusters that contain the 2 farthest pairs of observations which mean they are not alike.
