Graphics Reference
In-Depth Information
Figure
.
.
Heatmap of the dentitio data using Manhattan distance and the average linkage method for
both columns and rows
herefore,hierarchical clustering can beavaluable tool for rearranging variables and
observations in a data set in order to highlight interesting patterns.
Partitioning Cluster Analysis
11.3
Partitioning cluster analysis is a complexity reduction technique. It projects data
from a multidimensional (and most oten metric) space to a single nominal vari-
able, the cluster membership. he goal is to either find homogeneous subgroups in
the data, which are ideally as different as possible from each other, or to impose ar-
tificial grouping on the data. In any case, we want to increase our understanding of
the data using a “divide and conquer” approach that partitions a large and potentially
complex data set into segments that are easier to understand or handle.
