Page 90 -

Exercises 77

Mc Intyre RM, Blashfield RK (1980) A Nearest-Centroid Technique for Evaluating the
Minimum-Variance Clustering Procedure. Multivar. Behavioral Research, 15:225-238.
Milligan GW (1996) Clustering Validation: Results and Implications for Applied Analyses.
In: Arabie P, Hubert LJ, De Soete G (eds) Clustering and Classification. World Scientific
Pub. Co., Singapore.
Pipes A, Hovanessian SA (1977) Matrix-Computer Methods in Engineering. Robert E.
Krieger Pub. Co., Malabar, Florida.
Rohlf FJ (1982) Single-Link Clustering Algorithms. In: Krishnaiah PR, Kanal LN (eds),
Handbook of Statistics vo1.2, North Holland Pub. Co., 267-284.
Samrnon Jr JW (1969) A Nonlinear Mapping for Data Structure Analysis. IEEE Tr Comp,
18: 401-409.
Siegel S, Castellan Jr NJ (1988) Nonparametric Statistics for the Behavioural Sciences. Mc
Graw Hill Book Co.
Spath H (1980) Cluster Analysis Algorithms. Ellis Horwood, Ltd., England.

Exercises

3.1 Determine the tree clustering solutions of the +Cross data (Cluster.xls), using the
UWGMA linkage rule with the city-block and Chebychev norms. Explain the results.

3.2 Determine the tree clustering solutions of the xCross data (Cluster.xls), using the
UWGMA linkage rule with the city-block and Chebychev norms. Explain the results,
comparing them with those obtained in the previous exercise.
3.3 Determine the tree clustering solutions of the Globular data (Cluster.xls), using the
UPGMA and UWGMA linkage rules with the Euclidian norm. Compare the results
with those obtained using complete linkage in section 3.3.

3.4 Determine the tree clustering solutions of the Filamentmy data (Cluster.xls), using the
Ward method with the city-block norm. Compare the results with those obtained using
single linkage in section 3.3.1.

3.5 Perform a factor analysis of the Food data and determine the K-means cluster solutions
for c=3,4 and 5 in a two-dimensional space. Compare the results with those obtained
by multidimensional scaling (section 3.4).

3.6 Repeat the experiments on the Food data described in section 3.4 using a city-block
norm. Compare the results with those obtained using Ward method.

3.7 Perform a factor analysis of the Cork Stoppers data and determine the k-means cluster
solution for c=3 in a two-dimcnsional space. Compare the cluster results with the
supervised classification. Which class is in least agreement with the cluster solution
and why?

3.8 Perform a factor analysis of the CTG data. Determine which k-means clustering
solutions most resemble the supervised classification in three classes, N, Sand P.