IMPROVING THE ACCURACY OF REGRESSION MODELS BY METHODS OF POLYNOMIAL APPROXIMATION
Rubrics: MECHANICAL ENGINEERING
Authors:
1.
Bryansk State Technical University
(Department of Automated Technological Systems, Associate Professor)
Bryansk, Bryansk, Russian Federation
Bryansk, Bryansk, Russian Federation
Type:
Article
Pages:
from 4
to 12
Status:
Published
Received:
01.03.2024
Accepted:
13.03.2024
Published:
26.04.2024
Subject area:
VAC 1.2.2 Математическое моделирование, численные методы и комплексы программ
UDK 303.724.32 Регрессионный анализ
BBK 22 Физико-математические науки
UDK 303.724.32 Регрессионный анализ
BBK 22 Физико-математические науки
Language:
Russian
Keywords:
analysis, methods, model linearization, approximation, interpolation, extrapolation
Abstract (English):
The study objective is to assess the influence of the approximation method on the accuracy of the regression model. The task is to increase the accuracy of regression models using polynomial approximation methods. Research methods: the least square method, the polynomial approximation method. The novelty of the work: dependencies for the regression analysis of a multifactor experiment using polynomial approximation is developed, the influence of the approximation method on the accuracy of the regression model is analyzed. The study results: as a result of applying the developed dependencies, the accuracy of single-factor regression equations increased by 2.4 times, two-factor equations - by 1.7 times. Conclusions: the use of polynomial approximation makes it possible to increase the accuracy of single-factor and multifactor regression models, which increases their adequacy, the significance of coefficients and expands the possibilities of their application.
The study objective is to assess the influence of the approximation method on the accuracy of the regression model. The task is to increase the accuracy of regression models using polynomial approximation methods. Research methods: the least square method, the polynomial approximation method. The novelty of the work: dependencies for the regression analysis of a multifactor experiment using polynomial approximation is developed, the influence of the approximation method on the accuracy of the regression model is analyzed. The study results: as a result of applying the developed dependencies, the accuracy of single-factor regression equations increased by 2.4 times, two-factor equations - by 1.7 times. Conclusions: the use of polynomial approximation makes it possible to increase the accuracy of single-factor and multifactor regression models, which increases their adequacy, the significance of coefficients and expands the possibilities of their application.
Keywords:
analysis, methods, model linearization, approximation, interpolation, extrapolation
analysis, methods, model linearization, approximation, interpolation, extrapolation
1. Steposhina SV, Fedonin ON, Syanov SYu. Algorithmization in engineering problems: textbook. Moscow; Vologda: Infra-Enzheneria; 2023.
2. Steposhina SV. Choice of rational conditions for rolling and diamond smoothing of internal cylindrical surfaces of parts made of structural alloy steels [dissertation]. [Moscow (RF)]; 2013.
3. Danilov AM, Garkina IA. Interpolation, approximation, optimization: analysis and synthesis of complex systems: monograph. Penza: PGUAS; 2014.
4. Lvovsky EN. Statistical methods for constructing empirical formulas. Moscow: Visshaya Skola; 1982.
