For non-clustering SOM maps, which are r-by-c matrices, the Manhattan distance has the usual meaning. But for a clustering SOM map that has shape 1-by-k, the Manhattan distance simplifies to the ...

This example illustrates the use of regression analysis in a simple random cluster sampling design. The data are from S rndal, Swenson, and Wretman (1992, p. 652). A total of 284 Swedish ...

For example, suppose you set k = 5 and so each data item falls into one of clusters ... Data items in clusters 0 and 1 are closer to each other than data items in clusters 0 and 3. Self-organizing ...