Robust evaluation of regression parameters. The fuzzy theory and other models
Abstract
Linear regression parameters based on fuzzy theory are compared with other statistical approaches. A new algorithm of a simple weighted least squares method, independent of a priori information, is proposed. The algorithm was verified on model data, and its adequacy was confirmed with the use of standard criteria. The algorithm has been implemented as Python language computer program. New method of calculation of the scatter of fuzzy dependent variable around its mediane value, as well as the upper and lower bonds of fuzzy regression equations have been developed and verified. Proposed methods are shown to be useful alternatives to the most popular methods for constructing linear regression, which assume a normal distribution of errors.
Downloads
References
Onizhuk M.O., Ivanov V.V., Panteleimonov A.V., Kholin Yu.V. Alternative Methods for Constructing of Linear Regressions. Method and object Chemical Analysis. 2017, 12(3), 105-111. https://doi.org/10.17721/moca.2017.105-111
Berdnyk M. I., Onizhuk M. O., Ivanov V. V. Methods for building linear regression equations in the “structure-property” problems. Kharkov University Bulletin. Chemical Series. 2018, 30 (53), 6-17. https://doi.org/10.26565/2220-637X-2018-30-01
Berdnyk M.I., Zakharov A.B., Ivanov V.V. Application of L1-regularization approach in QSAR problem. Linear regression and artificial neural networks. Method and Object Chemical Analysis. 2019, 14(2), 79-90. https://doi.org/10.17721/moca.2019.79-90
Zakharov A.B., Dyachenko A.V., Ivanov V.V. Topological Characteristics of Iterated Line Graphs in QSAR Problem: Octane Numbers of Saturated Hydrocarbons. Journal of Chemometrics. 2019, 33 (9), e3169. https://doi.org/10.1002/cem.3169
Zakharov A. B., Tsarenko D. K., Ivanov V. V. Topological characteristics of iterated line graphs in the QSAR problem: a multigraph in the description of properties of unsaturated hydrocarbons. Struct Chem. 2021, 32, 1629-1639. https://doi.org/10.1002/cem.3169
Fuzzy logic in Chemistry. Rouvray D.H. ed. Academic Press, London, 1997, 364 p.
Zadeh L. A. The Concept of a Linguistic Variable and its Application to Approximate Reasoning-I. Information Sciences. 1975, 8, 199-249. https://doi.org/10.1016/0020-0255(75)90036-5
Dubois D., Prade H. Possibility Theory, Probability Theory and Multiple-Valued Logics: A Clarification. Annals of Mathematics and Artificial Intelligence. 2001, 32, 35–66. https://doi.org/10.1023/A:1016740830286
Bloomfield P., Steiger W.L. Least Absolute Deviations. Theory, Applications and Algorithms. Boston: Birkhäuser, 1983, 351 p.
Hanss M. Applied Fuzzy Arithmetic. An Introduction with Engineering Applications. Springer-Verlag Berlin Heidelberg 2005, 256 p.
Tanaka H., Uegima S., Asai K. Linear Regression Analysis with Fuzzy Model. IEEE Trans. on Systems, Man and Cybernetics. 1982,12, 903–907. http://dx.doi.org/10.1109/TSMC.1982.4308925
Diamond P. Fuzzy Least Squares. Information Sciences. 1988, 46, 141–157. https://doi.org/10.1016/0020-0255(88)90047-3
de Andrés-Sánchez J. Fuzzy Regression Analysis: An Actuarial Perspective in Fuzzy Statistical Decision-Making Theory and Applications. Springer. 2016. 173-201. https://doi.org/10.1007/978-3-319-39014-7_11
Conventional and fuzzy regression theory and engineering applications. Hrissanthou V., Spiliotis M. (eds). Nova Science Publishers Inc, New York. 2018, 332 p.
Haggag M.M.M. A New Fuzzy Regression Model by Mixing Fuzzy and Crisp Inputs. American Review of Mathematics and Statistics. 2018, 6(2), 9-25. https://doi.org/10.15640/arms.v6n2a2
Golbraikh A., Tropsha A. Beware of Q2! Journal of Molecular Graphics and Modelling. 2002, 20, 269-276. https://doi.org/10.1016/S1093-3263(01)00123-1
Alexander D.L.J., Tropsha A., Winkler D.A. Beware of R2: Simple, Unambiguous Assessment of the Prediction Accuracy of QSAR and QSPR Models. J. Chem. Inf. Model. 2015, 55 (7), 1316–1322. https://doi.org/10.1021/acs.jcim.5b00206
Huber P.J. Robust Statistics. J. Wiley and sons, New York. 1981, 308 p.
Kholin Yu. V. A quantitative physicochemical analysis of complexation in solutions and on the surface of complexing silicas: meaningful models, mathematical methods and their application. Kharkiv, Folio, 2000, 294 p. (in Rus).
