Page 91 -

78 3 Data Clustering

3.9 Repeat the experiment of the previous exercise for the other values of c between 2 and 5.
Compute the cluster merit factors and determine the most adequate clustcr solutions.

3.10 Perform a replication analysis for thc clustering solution of the Cork Stoppers data with
c=3.

3.1 1 Apply multidi~nensional scaling to the Rocks dataset and compare with the results
obtained using factor analysis (scction 3.5). Perform k-means clustering and assess
with cluster merit factors what are the most adequate solutions.
3.12 Determine the tree clustering solutions of the Rocks data in a two-dimensional factor
space, using single and complete linkage with the Euclidian norm. Compare the results
with those obtained in section 3.5.