Species Associations and Conditional Clustering: Clustering With or Without Pairwise Resemblances
Traditional procedures for clustering objects consist of two steps: measuring pairwise resemblance based on the attributes, and a clustering algorithm. The use of pairwise resemblances can be avoided; a set of objects can be represented as a set of lists of attribute states; an application of the Laplace indifference principle then allows an estimate to be made of the probability of each list as representative of an association of objects. By use of set-covering procedures, the object associations having maximum joint probability are found. The procedure is generalized to multistate unordered and ordered attributes, to frequencies, and to directly obtained relational data.
KeywordsOptimal Covering Species Association Initial Pair Missing Element Fortran Subroutine
Unable to display preview. Download preview PDF.
- Andersen, E.B. 1980. Discrete statistical models with social science applications. North-Holland, Amsterdam.Google Scholar
- André, H.M. 1984. Overlapping recurrent groups: an extension of Fager’s concept and algorithm. Biometrié-Praximétrie 24: 49–65.Google Scholar
- Dale, M.B. 1971. Information analysis of quantitative data. Statistical Ecology 3: 133–148.Google Scholar
- Fine, T.L. 1973. Theories of probability. Academic, New York.Google Scholar
- Garfinkel, R., and G.L. Nemhauser, 1972. Integer programming, Wiley, New York.Google Scholar
- Rasch, G. 1960. Probabilistic models for some intelligence and attainment tests. Danmarks Paedagogistic Institut, Copenhagen.Google Scholar
- Shore, J.E., and R.W. Johnson, 1980. Axiomatic derivation of the principle of maximum entropy and the principle of minimum cross entropy. IEEE Trans. Inform. Theory IT-26:26–37.Google Scholar
- Tjur, T. 1982. A connection between Rasch’s item analysis model and a multiplicative Poisson model. Scand. J. Statist. 9: 23–30.Google Scholar