“…In this paper we are interested in ML methods with some mechanisms that can be analogous to the utility functions defined in RUM. The combination of MNL with radial basis functions is known in the ML community as Kernel Logistic Regression (KLR) (Zhu and Hastie, 2005;Cawley and Talbot, 2005;Maalouf and Trafalis, 2011;Liu et al, 2016;Ouyed and Allili, 2018;Martín-Baos et al, 2020). The motivation underlying the use of this method, as opposed to other prominent ML methods such as RF, is that it allows for the identification of utility functions.…”
Section: Literature Reviewmentioning
confidence: 99%
“…A preliminary work from the authors (Martín-Baos et al, 2020) suggests the use of the KLR method as a promising tool in modelling individual behaviour. KLR has potential use in a constellation of disciplines such as marketing research, social sciences, health economics, among others.…”
The success of machine-learning methods is spreading their use to many different fields. This paper analyses one of these methods, the Kernel Logistic Regression (KLR), from the point of view of Random Utility Model (RUM) and proposes the use of the KLR to specify the utilities in RUM, freeing the modeller from the need to postulate a functional relation between the features. A Monte Carlo simulation study is conducted to empirically compare KLR with the Multinomial Logit (MNL) method, the Support Vector Machine (SVM) and the Random Forests (RF). We have shown that, using simulated data, KLR is the only method that achieves maximum accuracy and leads to an unbiased willingness-to-pay estimator for non-linear phenomena. In a real travel mode choice problem, RF achieved the highest predictive accuracy, followed by KLR. However, KLR allows for the calculation of indicators such as the value of time, which is of great importance in the context of transportation.
“…In this paper we are interested in ML methods with some mechanisms that can be analogous to the utility functions defined in RUM. The combination of MNL with radial basis functions is known in the ML community as Kernel Logistic Regression (KLR) (Zhu and Hastie, 2005;Cawley and Talbot, 2005;Maalouf and Trafalis, 2011;Liu et al, 2016;Ouyed and Allili, 2018;Martín-Baos et al, 2020). The motivation underlying the use of this method, as opposed to other prominent ML methods such as RF, is that it allows for the identification of utility functions.…”
Section: Literature Reviewmentioning
confidence: 99%
“…A preliminary work from the authors (Martín-Baos et al, 2020) suggests the use of the KLR method as a promising tool in modelling individual behaviour. KLR has potential use in a constellation of disciplines such as marketing research, social sciences, health economics, among others.…”
The success of machine-learning methods is spreading their use to many different fields. This paper analyses one of these methods, the Kernel Logistic Regression (KLR), from the point of view of Random Utility Model (RUM) and proposes the use of the KLR to specify the utilities in RUM, freeing the modeller from the need to postulate a functional relation between the features. A Monte Carlo simulation study is conducted to empirically compare KLR with the Multinomial Logit (MNL) method, the Support Vector Machine (SVM) and the Random Forests (RF). We have shown that, using simulated data, KLR is the only method that achieves maximum accuracy and leads to an unbiased willingness-to-pay estimator for non-linear phenomena. In a real travel mode choice problem, RF achieved the highest predictive accuracy, followed by KLR. However, KLR allows for the calculation of indicators such as the value of time, which is of great importance in the context of transportation.
“…. Variabels in this research consisted of educational background (ED), employment (EM) as independent variables and digital mobile scanner usage (DMS) as dependent variables [15]. About 340 respondents accross several big cities in Indonesia from various backgrounds participated in this research but reduced to 310 due to lack of completed informations [16].…”
As of today, the mobile apps may be downloaded everywhere. The development of mobile apps depends on the type of the work. An increasing use of mobile app is scanner apps due to an easy use. This paper presents the regression analysis on employment and educational background of the mobile scanner app because this research used category in the questionnaire. The use of logistic regression is to prove that any different comparisons are detected between employment and educational background so that the use of mobile scanner can be optimally used. The results show that educational background and employment have vital roles for mobile scanner adoption. This study also proves that previous researches on mobile scanner adoption were true for UTAUT model and comparison analysis.
“…Для визначення параметрів та структури моделей типу (7) можна використовувати різні методи структурної ідентифікації [31][32][33][34][35].…”
Section: метод структурної ідентифікації моделейunclassified
“…Задача розподілу даних на пробну та контрольну підвибірки не є тривіальною [34]. Лише у випадку, коли точно визначені закони розподілу похибок та апріорно виконано ряд інших припущень, можна вирішити цю задачу аналітично.…”
Section: метод структурної ідентифікації моделейunclassified
The paper considers a group of polynomial models of various characteristics of an optical fiber (OF) depending on the wavelength and chemical composition of the fiber. A method for structural identification of such models is proposed. The following characteristics are considered: the refractive index of the fiber core and cladding, group refractive index, group velocity, dispersion coefficients, numerical aperture, cutoff wavelength of the fundamental mode, etc. An analysis of the well-known Cauchy, Lorentz-Lorenz equations, Sellmeier’s formulas, etc. is given in relation to the problem being solved. The applied method of structural identification provides for the decomposition of a complex computational problem into simpler ones. This technique involves the identification of polynomial models for different samples of a substance. After that, structural identification is performed by the parameter of the additives to quartz glass. The proposed method and models are tested on the example of parameter values: the wavelength range is from 0.8 to 1.8 μm, the type of optical fiber is single-mode, and the refractive index is stepped. For calculations, the tabular values of the coefficients of the Sellmeier formula for SiO2 with GeO2 additions from 0% to 13.5% were used. It is shown that the dependence of the main characteristics of OF on wavelength and chemical composition is modeled with sufficient accuracy by a polynomial model. Indicators of the highest degree on two arguments can be limited to the third degree. The synthesized models have an interpolation and extrapolation error in the considered ranges of the order of 0.001%. This makes it possible to recommend them for scientific and engineering applications, as well as for solving problems of the production of organic matter with predictable characteristics.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.