21 Logical functions in the problems of empirical prediction

Publisher Summary An empirical table is considered whose elements are results of measuring a number of features of a subset of objects chosen from a general population Γ. This chapter discusses algorithms for solving problems in the case of empirical tables characterized by either a great number of features, a small size of sample or by heterogeneity of features (gaps are possible in tables for both cases). The table for which at least one of these properties is fulfilled is called RP-tables. Such tables are common in complex studies in medicine, sociology and geology. The chapter discusses requirements for a class of decision rules. The quality of a decision rule of recognition is known to depend on the information content of the initial system features, on the choice of the class of decision rules, on the size and representation of sample and optimization procedure of this criterion as well.