Premium
Implication operators versus regression analysis: Application to time series data
Author(s) -
Shnaider Eliahu,
Lynch Tim,
Bandler Wyllis
Publication year - 1993
Publication title -
international journal of intelligent systems
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.291
H-Index - 87
eISSN - 1098-111X
pISSN - 0884-8173
DOI - 10.1002/int.4550080902
Subject(s) - unavailability , computer science , operator (biology) , ordinary least squares , consistency (knowledge bases) , regression , linear regression , contrast (vision) , series (stratigraphy) , regression analysis , range (aeronautics) , intersection (aeronautics) , regression diagnostic , design matrix , data mining , mathematics , statistics , polynomial regression , machine learning , artificial intelligence , paleontology , biology , biochemistry , chemistry , materials science , repressor , transcription factor , engineering , composite material , gene , aerospace engineering
This article constitutes a study which compares the performance of implication operators to that of linear regression (Ordinary Least Squares—OLS) method. the emphasis of this study is focused on advantages versus disadvantages of each method from the standpoint of practical application of the method. Thus both methods are applied to solve identical problems, and then both—the process of application as well as the consistency of results—are studied. For the purpose of testing the consistency of results, both methods are applied to several ranges of the time series data. The ranges of the time series are such that most of the observations are identical. In other words, the intersection of the three time ranges covers most of these ranges, and thus the variation from one time range to another is slight. As the research progressed, it became apparent that conceptual and theoretical advantages of implication operator methodologies over the regression methods Some very restrictive conditions are required for an appropriate application of regression methods 1–3 but not for the use of implication operators 4–7. are countered by the incompleteness of research in that area (of implication operators) as reflected by the unavailability of significance indicators. Thus, both methods have some advantages and some disadvantages. the methodology of implication operators has a potential to become the superior method of the two, Because it does not require restrictions on the applications of the implication operators methods in contrast to the regression methods. once the research and development of these techniques is complete, and the problem of significance tests is solved. But until then, the preferred approach is to utilize both methods to extract maximum information about phenomena under investigation, and then to apply common sense to sort out and interpret the results. As a greater number of relevant but independent methods of analysis are applied to the investigation process, the results could either reinforce one another‐if they at least approximately point in the same direction‐or call for some caution in the case of contradictory outcome. © 1993 John Wiley & Sons, Inc.