400
3
Conclusions
C. Barcelo· V. Pawlowsky . G. Bohling
Interpretation of results from a cluster-analysis procedure is never the result of
merely looking at some numbers. Groups have to be reasonable in our case from
a geological point of view, and thus the opinion of a geologist is lacking at this
point. Nevertheless, natural groups that are clearly different will be quite stable
if subjected to different clustering methods. This is not the case of the Darss Sill
data set, where the three different approaches presented clearly lead to different
groupings of data. In the ternary diagram of Fig. 3, the data seem to follow a fairly continous distribution. This fact suggests that the data might be better treated
with a continuous approach, possibly in an appropiate transformed space. In
such a case, available methods have to be carefully checked for underlying hypotheses, particularly concerning assumed models. The necessity of a more profound study of available information is a must in the present case.
References
Aitchison, J. 1986, The Statistical Analysis of Compositional Data: Chapmann and Hall, London (GB),416p.
Barcelo, e., Pawlowsky, V., and Grunsky, E., 1996, Some aspects of transformations of compositional data and the identification of outliers: Mathematical Geology, vol. 28, no. 4,
pp.501-518.
Bohling, G. e., Davis, J. e., Olea, R. A., and Harff, J., 1998, Singularity and nonnormality in
the classification of compositional data: submitted to Mathematical Geology, vol. 30,
no. 1, pp.5-20.
Davis, J. C., Harff, J., Lemke, W., Olea, R. M., Tauber, E, Bohling, G.C., and Zhou Di, 1995,
Analysis of sedimentary facies by regionalized classification: Kansas Geological Survey
Open File Report 95-4, 28 p.
Everitt, B.S., 1993, Cluster Analysis: Edward Arnold, New York (USA), 3rd ed., 170 p.
Harff, J., and Davis, J.e. 1990, Regionalization in geology by multivariate classification:
Mathematical Geology, vol. 22, no. 5, pp. 573-588.
McLachlan, G. J., and Basford, K.E.1988,Mixture models. Inference and applications to clustering: Marcel Dekker, New York (USA), 253 p.
Sampson, R. J. 1988, Surface III® User's Manual, Kansas Geological Survey.
Statistical Sciences 1995, S-PLUS® Guide to Statistical and Mathematical Analysis, Version
3.3, Seattle: StatSci, a division of MathSoft, Inc.
3
Conclusions
C. Barcelo· V. Pawlowsky . G. Bohling
Interpretation of results from a cluster-analysis procedure is never the result of
merely looking at some numbers. Groups have to be reasonable in our case from
a geological point of view, and thus the opinion of a geologist is lacking at this
point. Nevertheless, natural groups that are clearly different will be quite stable
if subjected to different clustering methods. This is not the case of the Darss Sill
data set, where the three different approaches presented clearly lead to different
groupings of data. In the ternary diagram of Fig. 3, the data seem to follow a fairly continous distribution. This fact suggests that the data might be better treated
with a continuous approach, possibly in an appropiate transformed space. In
such a case, available methods have to be carefully checked for underlying hypotheses, particularly concerning assumed models. The necessity of a more profound study of available information is a must in the present case.
References
Aitchison, J. 1986, The Statistical Analysis of Compositional Data: Chapmann and Hall, London (GB),416p.
Barcelo, e., Pawlowsky, V., and Grunsky, E., 1996, Some aspects of transformations of compositional data and the identification of outliers: Mathematical Geology, vol. 28, no. 4,
pp.501-518.
Bohling, G. e., Davis, J. e., Olea, R. A., and Harff, J., 1998, Singularity and nonnormality in
the classification of compositional data: submitted to Mathematical Geology, vol. 30,
no. 1, pp.5-20.
Davis, J. C., Harff, J., Lemke, W., Olea, R. M., Tauber, E, Bohling, G.C., and Zhou Di, 1995,
Analysis of sedimentary facies by regionalized classification: Kansas Geological Survey
Open File Report 95-4, 28 p.
Everitt, B.S., 1993, Cluster Analysis: Edward Arnold, New York (USA), 3rd ed., 170 p.
Harff, J., and Davis, J.e. 1990, Regionalization in geology by multivariate classification:
Mathematical Geology, vol. 22, no. 5, pp. 573-588.
McLachlan, G. J., and Basford, K.E.1988,Mixture models. Inference and applications to clustering: Marcel Dekker, New York (USA), 253 p.
Sampson, R. J. 1988, Surface III® User's Manual, Kansas Geological Survey.
Statistical Sciences 1995, S-PLUS® Guide to Statistical and Mathematical Analysis, Version
3.3, Seattle: StatSci, a division of MathSoft, Inc.