MULTIVARIATE STATISTICAL ANALYSIS OF ATMOSPHERIC PROCESSES

摘要

Discriminant function analysis, described in Section I (James and Shorr), is an objective method for classifying sets of observed weather elements into groups, the classification de?pending upon, say, events observed on the next day. Probabi1ities can be derived for various classes of events, and a screening type procedure can be applied to select the important classificatory variables. Many meteorological problems are concerned with prediction of “classes” (say, above or below normal) rather than the specific value of the variable.nIn Section III (Ostby and Verigas) the design of a large-scale experiment to examine this problem is described.