Some generalization of combined method “factor analyze + taxonomy”

The article discusses the generalized method of analysis of observational data needed to detect regularities assess the importance of attributes of the objects and finding the hidden factors. The proposed method is based on the use of discriminated analysis, taxonomy and evaluation of the information subsystems signs. Analysis of the data suggests not only the search for the underlying factors, but also the discovery of hidden patterns. A model of the problem of detection of regularities is suggested, that is its reduction to the discriminanted analysis - the problem of separating sets. In this case, it was sufficient to consider the case of the partition into two classes. In the absence of an analytical description of this partition, work with objects, you can only interact with their devices and examinations. Then the classes recovery makes by the relevant case sets. Specific applications of the results are the search problem patterns in non-formalized problems of mathematical economics, mathematical biology and medicine. The article shows that the theory of algorithms allows us to analyze not only formalized tasks, but also, in principle, non-formalizable.