Nonlinear Regression Estimation Research Articles

Data-driven strategies for accurate volume-to-biomass conversions

ABSTRACT This document presents a method for estimating conversion factors from volume to green weight and from volume to oven dry weight using ratio and linear and nonlinear regression estimators, and the results of its use in nine raw materials, including roundwood from harvest and sawmill offcuts. Although no differences were detected in either the value or accuracy of the linear factors, the linear estimator tended to have slightly higher accuracy than the ratio estimator. Linear conversion factors obtained using regression through the origin were accurate, especially those for roundwood. In general, the conversion factors from volume to green weight were more precise than those from volume to dry weight. Conversion factors from volume to dry weight are useful only in situations in which the moisture range of application is like that used for obtaining those average factors. In addition, given that the green volume-weight factor can be adjusted to the observed moisture in the raw material for which dry weight is intended to be determined, it is advisable to concentrate efforts on this type of factor in future studies. Nonlinear regression estimators are an alternative to linear regression estimators through the origin for determining conversion factors from volume to dry weight. In all the analyzed raw materials, a decrease in the conversion factor from volume to dry weight was observed as the moisture content increased. This is an unexpected outcome that cannot be explained by the variables recorded in this study.

Tree-based boosting with functional data

In this article we propose a boosting algorithm for regression with functional explanatory variables and scalar responses. The algorithm uses decision trees constructed with multiple projections as the “base-learners”, which we call “functional multi-index trees”. We establish identifiability conditions for these trees and introduce two algorithms to compute them. We use numerical experiments to investigate the performance of our method and compare it with several linear and nonlinear regression estimators, including recently proposed nonparametric and semiparametric functional additive estimators. Simulation studies show that the proposed method is consistently among the top performers, whereas the performance of existing alternatives can vary substantially across different settings. In a real example, we apply our method to predict electricity demand using price curves and show that our estimator provides better predictions compared to its competitors, especially when one adjusts for seasonality.

Inflation and Its Uncertainty: Evidence from Indonesia and the Philippines

This article investigates the relationship between inflation and its uncertainty under the inflation-targeting regime in Indonesia and the Philippines over 2005(7)–2021(12). The Kernel density function and entropy index applications reveal that the price level in both countries is heterogeneous. The non-linear quantile regression estimation shows that inflation affects inflation uncertainty positively. Inflation more likely induces its uncertainty in higher quantiles. The effect exhibits an inconsistent linearity property. While the symmetric behavior holds in the Philippines, inflation in the case of Indonesia affects the inflation uncertainty differently, not only inter-quantiles but also intra-quantiles. Given that Indonesia and the Philippines have low inflation rates relative to other emerging markets, our findings suggest that announcing higher inflation targets may be costly to provoke higher inflation uncertainty. Accordingly, controlling inflation uncertainty in the inflation-targeting regime can be achieved if monetary policy is credible and independent. JEL Classification C14, C21, D08, E31, O23

Prediction of stiffness modulus of bituminous mixtures using the applications of multi-expression programming and gene expression programming

An essential factor in the design and in situ performance of asphalt concrete pavements is its elastic-stiffness properties. The objective of the present study, is to develop empirical equations to estimate indirect tensile stiffness modulus (ITSM) of bituminous mixture using the soft computing techniques of multi expression programming (MEP) and gene expression programming (GEP). The soft computing models were developed using experimental data points containing laboratory and field blended dense-graded hot mix asphalt (HMA) and gap-graded stone matrix asphalt (SMA) mixtures including three asphalt binders. The dominant parameters covered mix physical and volumetric properties, test temperatures, and the mix variables needed during quality control and assurance assessments. The soft computing models could successfully estimate the stiffness modulus with notable values of R values between 0.835 and 0.944. The GEP equations estimated the elastic stiffness modulus with a higher rate of accuracy when compared to the MEP, non-linear and linear regression estimates.

Forecasting Selected Colombian Shares Using a Hybrid ARIMA-SVR Model

Forecasting future values of Colombian companies traded on the New York Stock Exchange is a daily challenge for investors, due to these stocks’ high volatility. There are several forecasting models for forecasting time series data, such as the autoregressive integrated moving average (ARIMA) model, which has been considered the most-used regression model in time series prediction for the last four decades, although the ARIMA model cannot estimate non-linear regression behavior caused by high volatility in the time series. In addition, the support vector regression (SVR) model is a pioneering machine learning approach for solving nonlinear regression estimation procedures. For this reason, this paper proposes using a hybrid model benefiting from ARIMA and support vector regression (SVR) models to forecast daily and cumulative returns of selected Colombian companies. For testing purposes, close prices of Bancolombia, Ecopetrol, Tecnoglass, and Grupo Aval were used; these are relevant Colombian organizations quoted on the New York Stock Exchange (NYSE).

Economic Growth and The Quality of Environment: Evidence of The Environmental Kuznets Curve (EKC) in Indonesia

This study proved the EKC hypothesis results of the trade-off between economic growth and environmental degradation in Indonesia. The researchers used the non-linear quadratic regression estimation. This study found that the EKC hypothesis was almost valid in all islands in Indonesia, such as Sumatra Island, Sulawesi Island, Borneo Island, and Small Island in the Eastern Region of Indonesia. Thus, the economic growth causes the environmental gradation while the trade-off did not apply in Java and Bali. Government policies regarding the environment and globalization played an important role in generating and transferring production technology that saved resources and used cleaner technology from developed countries to developing countries so the EKC hypothesis does not apply in Indonesia.

SARS-CoV-2 Spike and Neutralizing Antibody Kinetics 90 Days after Three Doses of BNT162b2 mRNA COVID-19 Vaccine in Singapore.

Background: We evaluated the post-booster (BNT162b2) antibody responses in Singapore. Methods: Participants (n = 43) were tested pre-booster and 20/30/60/90 days post-booster. Participants were boosted 120–240 days (mean 214 days) after their second dose and had no history or serologic evidence of prior COVID-19 infection; all participants had undetectable SARS-CoV-2 nucleocapsid antibodies throughout the study. Total nucleocapsid and spike antibodies (S-Ab) were assessed on the Roche Elecsys e802 and neutralizing antibody (N-Ab) on the Snibe quantitative N-Ab assay. Results: Pre-booster median S-Ab/N-Ab titers were 829 BAU/mL/0.83 µg/mL; 2 participants were below manufacturer’s N-Ab cut-offs of 0.3 µg/mL (0.192 and 0.229). Both S-Ab and N-Ab titers peaked at 30 days post-booster (median S-Ab 25,220 BAU/mL and N-Ab 30.3 µg/mL) at 30–37× pre-booster median levels. These peak post-booster S-Ab/N-Ab titers were 11× (25,220 vs. 2235 BAU/mL) and 9× (30.3 vs. 3.52 µg/mL) higher than the previously reported peak post-second dose levels. Antibody titers declined to 12,315 BAU/mL (51% decrease) and 14.3 µg/mL (53% decrease) 90 days post-booster. Non-linear regression estimates for S-Ab/N-Ab half-lives were 44/58 days. At 180 days post-booster, S-Ab/N-Ab are estimated to be 2671 BAU/mL/4.83 µg/mL. Conclusions: Both S-Ab and N-Ab show a good response following post-booster vaccination, with half-lives that may provide a prolonged antibody response.

Validation and test\u2013retest repeatability performance of parametric methods for [11C]UCB-J PET

[11C]UCB-J is a PET radioligand that binds to the presynaptic vesicle glycoprotein 2A. Therefore, [11C]UCB-J PET may serve as an in vivo marker of synaptic integrity. The main objective of this study was to evaluate the quantitative accuracy and the 28-day test–retest repeatability (TRT) of various parametric quantitative methods for dynamic [11C]UCB-J studies in Alzheimer’s disease (AD) patients and healthy controls (HC). Eight HCs and seven AD patients underwent two 60-min dynamic [11C]UCB-J PET scans with arterial sampling over a 28-day interval. Several plasma-input based and reference-region based parametric methods were used to generate parametric images using metabolite corrected plasma activity as input function or white matter semi-ovale as reference region. Different parametric outcomes were compared regionally with corresponding non-linear regression (NLR) estimates. Furthermore, the 28-day TRT was assessed for all parametric methods. Spectral analysis (SA) and Logan graphical analysis showed high correlations with NLR estimates. Receptor parametric mapping (RPM) and simplified reference tissue model 2 (SRTM2) BPND, and reference Logan (RLogan) distribution volume ratio (DVR) regional estimates correlated well with plasma-input derived DVR and SRTM BPND. Among the multilinear reference tissue model (MRTM) methods, MRTM1 had the best correspondence with DVR and SRTM BPND. Among the parametric methods evaluated, spectral analysis (SA) and SRTM2 were the best plasma-input and reference tissue methods, respectively, to obtain quantitatively accurate and repeatable parametric images for dynamic [11C]UCB-J PET.

Thin nanocrystalline semiconductor films as selective chemical sensors for ammonia, acetone and other donors

Studied are the films of variously doped polycrystalline n-semiconductors (ZnO, SnO2) as selective sensitive elements (SE) of chemical sensors for various gases and vapours (ammonia, acetone, propane, ethanol, hexane, solvent, turpentine etc.) in artificial air. It has been revealed that their conductivity changes under temperature modulation makes possible data processing which identifies the impurities above. This processing is based on nonlinear regression estimation of so called principal parameters which set is unique for every concentration of every of the gases/vapours.

Modeling Alleviative Effects of Ca, Mg, and K on Cu-Induced Oxidative Stress in Grapevine Roots Grown Hydroponically.

The aim of this study was to determine the pattern of alleviation effects of calcium (Ca), magnesium (Mg), and potassium (K) on copper (Cu)-induced oxidative toxicity in grapevine roots. Root growth, Cu and cation accumulation, reactive oxygen species (ROS) production, and antioxidant activities were examined in grapevine roots grown in nutrient solutions. The experimental setting was divided into three sets; each set contained a check (Hoagland solution only) and four treatments of simultaneous exposure to 15 μM Cu with four cation levels (i.e., Ca set: 0.5, 2.5, 5, and 10 mM Ca; Mg set: 0.2, 2, 4, and 8 mM Mg; K set: 0.6, 2.4, 4.8, and 9.6 mM K). A damage assessment model (DAM)-based approach was then developed to construct the dose-effect relationship between cation levels and the alleviation effects on Cu-induced oxidative stress. Model parameterization was performed by fitting the model to the experimental data using a nonlinear regression estimation. All data were analyzed by a one-way analysis of variance (ANOVA), followed by multiple comparisons using the least significant difference (LSD) test. The results showed that significant inhibitory effects on the elongation of roots occurred in grapevine roots treated with 15 μM Cu. The addition of Ca and Mg significantly mitigated phytotoxicity in root growth, whereas no significant effect of K treatment on root growth was found. With respect to oxidative stress, ROS and malondialdehyde (MDA) contents, as well as antioxidant (superoxide dismutase (SOD), catalase (CAT), and ascorbate peroxidase (APX)) activities, were stimulated in the roots after exposure to 15 μM Cu for three days. Moreover, H2O2 levels decreased significantly as Ca, Mg, and K concentrations increased, indicating that the coexistence of these cations effectively alleviated Cu-induced oxidative stress; however, alleviative effects were not observed in the assessment of the MDA content and antioxidant enzyme activities. Based on the DAM, an exponential decay equation was developed and successfully applied to characterize the alleviative effects of Ca, Mg, and K on the H2O2 content induced by Cu in the roots. In addition, compared with Mg and K, Ca was the most effective cation in the alleviation of Cu-induced ROS. Based on the results, it could be concluded that Cu inhibited root growth and Ca and Mg absorption in grapevines, and stimulated the production of ROS, lipid peroxidation, and antioxidant enzymes. Furthermore, the alleviation effects of cations on Cu-induced ROS were well described by the DAM-based approach developed in the present study.

Perceptibility and Acceptability of Surface Gloss Variation Under Different Illuminants

The purpose of this study was to evaluate the influence of different illuminants on the perceptibility and acceptability of surface gloss variations and to determine limiting values. Eight composite resin specimens and one human tooth specimen were polished to obtain composite resin specimens with different gloss units (GU) of 10, 20, 30, 40, 50, 60, 70, and 80 and a human tooth specimen of 80 GU. Sixty observers compared the surface gloss of the specimens in a light booth. For perceptibility testing, the specimens were randomly positioned two at a time. The acceptability of the gloss variation was determined by comparing the composite resin specimens with the tooth specimen. The observers answered specific questions to determine the level and perceptibility and acceptability limits of gloss variations. All analysis was done with two illuminants (D65 and fluorescent light) used randomly. Data were submitted to a nonlinear probit model and nonlinear probit regression estimation (α=0.05). Significant differences in illuminants were observed for perceptibility (p<0.001) and acceptability (p=0.045). The perceptibility limit for D65 was 7.0 GU and 6.8 GU for fluorescent illuminant. The acceptability limit for D65 was 34.2 GU and 37.1 GU for fluorescent illuminant. More accurate perceptibility and acceptability judgments of the surface gloss of composite resin were made when the specimens were illuminated with D65 light.

Disparities in COVID-19 mortality by county racial composition and the role of spring social distancing measures

Available COVID-19 data shows higher shares of cases and deaths occur among Black Americans, but reporting of data by race is poor. This paper investigates disparities in county-level mortality rates across counties with higher and lower than national average Black population shares using nonlinear regression decomposition and estimates potential differential impact of social distancing measures. I find counties with Black population shares above the national share have mortality rates 2 to 3 times higher than in other counties. Observable differences in living conditions, health, and work characteristics reduce the disparity to approximately 1.25 to 1.65 overall, and explain 100% of the disparity at 21 days after the first case. Though higher rates of comorbidities in counties with higher Black population shares are an important predictor, living situation factors like single parenthood and population density are just as important. Higher rates of co-residence with grandchildren explain 11% of the 21 day disparity but do not appear important by 42 days, suggesting families may have been better able to protect vulnerable family members later in the epidemic. To analyze differential effects of social distancing measures use two approaches. First, I exploit the timing of interventions relative to the first case among counties that began their epidemic at the same time. Second, I use event study analysis to analyze within-county changes in mortality. Findings for social distancing measures are not always consistent across approaches. Overall, I find no evidence that school closures were less effective in counties with larger Black population shares, and some estimates suggest closures may have disproportionately helped more diverse counties and counties with high rates of grandparent and grandchild co-residence. Conversely, stay at home orders are less clearly associated with mortality in any counties, reaching peak unemployment did not reduce mortality in any models, and some estimates indicate reaching peak unemployment before the first case was associated with higher mortality rates, especially in more diverse counties.

Government Size, the Composition of Public Spending and Economic Growth in Netherland

The performance of the fiscal policy is influenced significantly by the relationships and associations among the governmental size, the composition public and private spending and the economic growth of a country. The primary goal of this research study is to evaluate the impact of these factors and to evaluate the significance of these factors in the economic growth of the Netherlands. The economy of Netherlands is characterized as the 17th largest in the world and it presents stable and sustainable growth. In this altercation the researcher aimed to evaluate the significance of the governmental spending and size. For this purpose, the researcher used the data from 2004-2014 from the 12 provinces of Netherlands. The data was subjected to a unit root analysis so that the stationarity properties of the panel data can be evaluated. The unit root test results showed that the variables were stationary at I(0). In order to abstain from the endogeneity issues that can be present in such types of datasets the researcher used level and per capita variables as a robustness evaluation. The empirical framework was based upon the Cobb-Douglas production function and used the modern CES substitution elasticities to compute as the inputs of private capital and government spending in the production function. The nonlinear least-squares regression estimation method was used to evaluate the impact of the variables upon one another. The results indicate that the public investments and current governmental disbursements are conjoined in order to account for the inflexibility of the communal budget. Moreover, the governmental spending was found to be greater than 85 percent indicating that the provincial sectors are focused upon the stimulation of the economy.

RECURRENT NEURAL NETWORKS AND NONLINEAR PREDICTION IN SUPPORT VECTOR MACHINES

The nonlinear regression estimation issues are solved by successful application of a novel neural network technique termed as support vector machines (SVMs). Evaluation of recurrent neural networks (RNNs) can assist in pattern recognition of several real-time applications and reduce the pattern mismatch. This paper provides a robust prediction model for multiple applications. Traditionally, back-propagation algorithms were used for training RNN. This paper predict system reliability by applying SVM learning algorithm to RNN. Comparison of the proposed model is done with the existing systems for analysis of prediction performance. These results indicate that the performance of proposed system exceeds that of the existing ones.

Perceptibility and Acceptability of Surface Gloss Variations in Dentistry.

To assess the visual perception of observers regarding perceptibility and acceptability of surface gloss variations of resin composites and to determine the limit of perceptibility and acceptability of gloss variations. Eight resin composite specimens and one human tooth specimen were fabricated. The resin specimens were polished to reach a surface gloss, in gloss units (GUs), of 10 GU, 20 GU, 30 GU, 40 GU, 50 GU, 60 GU, 70 GU, and 80 GU, and the human tooth specimen had a surface gloss of 80 GU. Sixty observers were selected to compare the surface gloss of the specimens in a light booth. For the perceptibility assessment, specimens were randomly displayed two at a time. Each observer performed a total of 144 observations. Observers answered two specific questions for determining the level and limit of perceptibility of gloss variations. The acceptability limit of gloss was determined by comparing the resin specimens with the tooth specimen. The observations were performed with dental practice scenarios (illuminant conditions, visualization field, and observers' education). Data were submitted to a nonlinear probit model and nonlinear regression estimation probit (5%). Differences in perceptibility and acceptability were observed for surface gloss variations (ΔGU) (p<0.001). Perceptibility increased with ΔGU (10<20<30<40<50=60=70), while acceptability decreased with ΔGU (0=10>20>30>40>50>60=70). Acceptability and perceptibility limits were 6.4 GU and 35.7 GU, respectively. Perceptibility and acceptability of surface gloss are influenced by gloss variations. A variation of 6.4 GU was required for 50% of observers to notice gloss variations. Concerning acceptability, observers required a variation of 35.7 GU to consider differences in gloss not acceptable.

List Experiments with Measurement Error

Measurement error threatens the validity of survey research, especially when studying sensitive questions. Although list experiments can help discourage deliberate misreporting, they may also suffer from nonstrategic measurement error due to flawed implementation and respondents’ inattention. Such error runs against the assumptions of the standard maximum likelihood regression (MLreg) estimator for list experiments and can result in misleading inferences, especially when the underlying sensitive trait is rare. We address this problem by providing new tools for diagnosing and mitigating measurement error in list experiments. First, we demonstrate that the nonlinear least squares regression (NLSreg) estimator proposed in Imai (2011) is robust to nonstrategic measurement error. Second, we offer a general model misspecification test to gauge the divergence of theMLregandNLSregestimates. Third, we show how to model measurement error directly, proposing new estimators that preserve the statistical efficiency ofMLregwhile improving robustness. Last, we revisit empirical studies shown to exhibit nonstrategic measurement error, and demonstrate that our tools readily diagnose and mitigate the bias. We conclude this article with a number of practical recommendations for applied researchers. The proposed methods are implemented through an open-source software package.

Allometric equations for predicting mineralomass in high-forest chestnut stands in Portugal

The assessment of nutrients in biomass tree-components is a time-consuming and expensive process, often involving tree felling, not always possible or desirable. Thus, mineralomass prediction equations are an important tool for the quantification of the nutrients exported in management and harvesting activities towards its replacement and sustainable management, as well as to evaluate the effect of other disturbances in the balance of ecosystems. Thus, given the importance of the relationship of biomass and nutrients (mineralomass) for dynamic and sustainable management of chestnut woodlands, above-ground mineralomass was studied in sweet chestnut (Castanea sativa Mill.) high-forest stands located in northern Portugal. Nutrient-specific prediction equations that allow estimating the mineralomass (N, P, K, Ca, Mg, S, B and C) stocked in the trees above the ground, considering the tree as a whole (stem + bark + branches + leaves + flowers) and seperately for each tree component: stem-wood, stem-bark, branches, leaves and flowers, based on tree dendrometric variables, DBH (diameter breast height) and total height, were developed.. Linear and non-linear regression estimation methods were used. Data analysis was based on information collected in destructive analysis of 34 felled trees, distributed by the existing diameter classes (10-65 cm) in three adult chestnut stands. Several linear and nonlinear equations were fitted by the least squares method to select models. A simultaneous fit by SUR method using iterative seemingly unrelated regression (ITSUR) was used for the final selected models. The best-fitting models are presented.

Predicting Stock Market Index Using Hybrid Intelligence Model

Predicting financial market changes is an important issue in time series analysis, receiving an increasing attention due to financial crisis. Autoregressive integrated moving average (ARIMA) model has been one of the most widely used linear models in time series forecasting but ARIMA model cannot capture nonlinear patterns easily. Generalized autoregressive conditional heteroscedasticity (GARCH) model applied understanding of volatility depending to the estimation of previous forecast error and current volatility, improving ARIMA model. Support vector machine (SVM) and artificial neural network (ANN) have been successfully applied in solving nonlinear regression estimation problems. This study proposes hybrid methodology that exploits unique strength of GARCH + SVM model, and GARCH + ANN model in forecasting stock index. Real data sets of stock prices FTSE Bursa Malaysia KLCI were used to examine the forecasting accuracy of the proposed model. The results shows that the proposed hybrid model achieves best forecasting compared to other model.

Somatic Cell Count and Milk Yield Losses in Goats

This work is aimed at prediction and quantification of goat milk yield (MY, kg) losses by the somatic cell count (SCC, 103.ml−1) in milk recording. The goal is to support the prevention of milk secretion disorders, milk yield and quality. During two years there were evaluated composition and properties of individual milk samples (n = 1 173). There were included animals of brown short‑haired (BSH) breed and BSH × white short‑haired in one flock. The linear and nonlinear regression, interpolation, extrapolation, approximation and qualified estimation were used for milk losses prediction along the SCC. The relevant values of the SCC arithmetic and geometric mean and median and MY arithmetic mean were: 1,400 103.ml−1 (with high variability of 128 %), 745 and 747 103.ml−1; 2.94 kg/ day. The relationship between fat and crude protein had a correlation coefficient of 0.395, P &lt; 0.001. There were stated the correlations between SCC and: lactose −0.416, P &lt; 0.001; solids non fat −0.25, P &lt; 0.001; MY −0.135, P &lt; 0.01. The relationship between SCC and MY was negative along months, 6 cases out of 7 and significant in March and May (P &lt; 0.05 and P &lt; 0.01). For instance predicted MY for SCC intervals 1–1.999, 2–2.999, 3–3.999, 4–4.999, 5–5.999, 6–6.999 and ≥ 7 thousands 103.ml−1 may be 3.99, 3.77, 3.6 , 3.46, 3.33, 3.23 and 3.13 kg in 3rd month etc. The corresponding MY losses by SCC are evident from this trend. The result use is focused on the original dairy goat farm and use elsewhere is possible in farms with similar SCC, goat milk yield and rearing system.

Nonparametric Regression Using Clusters

We present a fundamentally unique method of nonparametric regression using clusters and test it against classically established methods. We compare two nonlinear regression estimation packages called ‘NNS’, Viole (NNS: nonlinear nonparametric statistics, 2016), and ‘np’, Hayfield and Racine (J Stat Softw 27(5):1–32, 2008), with the help of a simulation using deterministic (DT) and stochastic (ST) regressor models. We find the respective coefficients of determination $$(R^2)$$ are close for DT models, while finding an advantage to NNS in ST and large sample cases. Regression coefficients are sometimes regarded as approximations to partial derivatives, especially in social sciences. Then, NNS alone has the ability to compute a range of partials evaluated at points within the sample and also out-of-sample. Thus NNS can provide a viable alternative to kernel based nonparametric regressions without using bandwidths for smoothing.