Using multivariate analysis on multielemental data.

Abstract

In the search for correlation between elements or among the analyzed samples, much time and effort can be devoted to the examination of multielemental data when the elements are taken two at a time, especially when the number of analyzed elements is relatively large. By representing the samples as points in an n-dimensional space, in which the coordinates are proportional to the elemental content, an overall view of all the available information may be obtained by visualization of the interrelationships among the points through several of the multivariate analysis methods that have been described. Such visualization is a useful step in the classification of samples into groups, particularly so when no a priori group criteria existed. The true significance of the grouping often becomes evident when other known information is superimposed on the multielemental analysis plots. Indeed, though it is not an essential part of the evaluation process, the importance of visualization of multiparameter data has been stressed most strongly(10), because in this way, the area for further intensive statistical analysis can be clearly demarcated.