ON THE REPRESENTATION OF ACTIVITY ARRAYS†

Abstract

A general methodology for empirical investigaiion is described which consists of: (i) the selection of a system on the object of investigation with respect to the purpose and constraints of investigation, (ii) the gathering of data for the system and organizing them in the form of activity arrays, (iii) the processing of the data through which certain time-invariant properties representing the data are determined, (iv) the interpretation of these time-invariant data representations from the standpoint of the purpose of investigation. It is shown that each activity array yields many different representations. A procedure through which the various representations can be obtained is described that consists of: (1) the selection of a mask specifying a pattern for sampling the data, (2) the sampling procedure through which certain types of time-invariant relations are determined for the given mask, (3) the simplification of the time-invariant relations, if necessary or desirable. A basis for the objective comparison of possible representations and ascertainment of the representation with the highest possible degree of determinism, subject to given constraints, is developed. The procedure is applicable to both well defined and fuzzy variables of any scale. No classification into input and output variables is required, although it is not prohibited.