Report of the Workshop on Current Problems and Future Developments in Multivariate Analysis

Abstract

Multivariate analysis is a scientific method in which the combined effect of many variables is studied. In statistics, the principal techniques of multivariate analysis are (multiple) regression, discriminant analysis, factor analysis and canonical correlation. In the earth sciences, special consideration must be given to the spatial settings of the rocks in a study area as well as to the dimension of geologic time. Multivariate mineral resource estimation should be performed according to the following steps. First the objective of the study is defined. This leads to the definition of a population from which one or more samples will be drawn. The objects in a sample have sizes, shapes, spatial arrangements and other attributes such as chemical composition allowing the definition of specific variables. This, in turn, leads to measurements. Each observation bears the values of many variables. The frequency distributions of the variables will be analysed in order to create new information as well as heuristics for solving the problems defined at the beginning of the study.