A SYSTEM FOR DEVELOPMENT OF MATHEMATICAL MODELS
Abstract
This paper presents a general quantitative system for association of theory and observation. The system, in formulation and presentation, is directed toward the needs of the user investigator. It is well organized for use with automatic machines in the computing and information processing. A multivariable distribution approach to model construction is used in accounting for errors and other sources of variation. As usual, hypothesized mathematical descriptions are modified in accordance with the observational data. The requirement of a state of control is regarded as fundamental. The question of unidentifiability is given prominent consideration; and in this connection, the general necessity of calibration is established and emphasized. The principle of maximum likelihood is suggested as the most acceptable ranking criterion for the system; but modifications or decision-theoretic extensions are not precluded. In short, the ordinary restrictive conditions are not imposed in this system. The fundamental ideas are discussed within the contextual framework of the system. The general principles for applying the system are presented and discussed. The most important classes of models are dealt with mathematically in detail.
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 23, 1964
- Accession Number
- AD0602733
Entities
People
- J. W. Hendricks
Organizations
- Naval Radiological Defense Laboratory