Answers are grounded in this paper and cite their sources.
Existing methods of examining the interrelationships among variables (R-mode) and among individual objects (Q-mode) demand measurements on an interval or ratio scale and are frequently not strictly applicable to many geomorphological studies where, for a variety of reasons, lower measuring scales are employed. By contrast, the information- and graph-theory methods use nominal scale data. Their use is illustrated by a discussion based on a 41 (objects) X 12 (variables) matrix relating to the cross-valley moraines of Baffin Island. The variables measure such factors as: moraine shape, location, till fabrics, and clast roundness. The information-theoretic approach generates a measure of distance (D) and computes the amount of shared information between the variables. Unconditional and conditional probability Tables are used to assess the stochastic dependence among states (i.e., small, medium, large) of two variables. Graphtheoretic methods are used to develop a measure of similarity that is then used in a classification of the moraines. The three characters that share most information are: two measures of clast roundness and the dip pattern of till fabric. Two classifications are developed that share considerable information with moraine shape and location within the valley. The study indicates that "hooked" and "asymmetric" moraines are quite different from each other. Since the techniques use nominal data, they should have general applicability in several fields of geology-specifically geomorphology, quaternary geology, and paleontology.
No takes yet. Share an insight, caveat, or question.
Andrews et al. (1971) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: