Accession Number : AD0715781

Title :   Numerical Taxonomy.

Descriptive Note : Technical rept.,

Corporate Author : STANFORD UNIV CALIF DEPT OF STATISTICS

Personal Author(s) : Solomon,Herbert

Report Date : 04 DEC 1970

Pagination or Media Count : 47

Abstract : Several recently developed clustering techniques for exploring multivariate data are presented and analyzed. These techniques run the gamut from evaluation of all partitions of an n-dimensional space to some step-wise procedures that inherently eliminate evaluation of many possible partitions. The former technique can produce optimum clusterings but is usually prohibitive in cost or impossible to manage on a computer. Both bases for clustering are employed on some specific data sets in the fields of linguistics, law and medicine. Comparisons are then made as to the efficiency of step-wise clustering schemes in arriving at meaningful clusterings for a reasonable cost. (Author)

Descriptors :   (*MULTIVARIATE ANALYSIS, STATISTICAL DATA), MATRICES(MATHEMATICS), DISPERSION RELATIONS, SET THEORY, RANDOM VARIABLES, CORRELATION TECHNIQUES, COMPUTER PROGRAMMING, ECONOMICS, SOCIAL SCIENCES

Subject Categories : Statistics and Probability

Distribution Statement : APPROVED FOR PUBLIC RELEASE