All MCQs

SOM aims to retain neighborhood relationships rather than exact geometry.
This makes the map useful for visual exploration.

Large neighboring distances can indicate cluster boundaries.
Small distances suggest homogeneous regions of the map.

Large early neighborhoods organize the whole map coherently.
Smaller later neighborhoods refine local prototypes.

Neighborhood updates encourage nearby grid units to represent similar regions.
This creates topological organization.

For each input, distances to prototype vectors are compared.
The nearest unit wins and guides the update.

SOM units compete to represent input vectors while preserving neighborhood structure.
The grid supports clustering and visualization.

A full distance matrix grows quadratically with the number of observations.
Memory and update costs limit naive implementations.

A mistaken early merge remains embedded in later clusters.
The algorithm's greedy nature affects the full hierarchy.

A horizontal cut separates the dendrogram into disconnected branches.
Each branch below the cut becomes a cluster.

A high-range variable can dominate pairwise distances.
Standardization gives variables more comparable influence.

Ward's method selects merges that cause the smallest loss of compactness.
It often produces relatively balanced, spherical clusters.

Average linkage balances information from all pairs between two groups.
It is less extreme than single or complete linkage.