Search

Featured Application: Modulus of elasticity for grain-supported carbonate rocks is successfully estimated by using simple estimation models. The methodology and models described in this paper can help engineers in the early design stages when they need a quick and easy estimate. The determination and estimation of elastic behaviour are essential in engineering practise, especially in quarrying, mining, construction, and all engineering professions that perform operations dealing with rock materials. Young's modulus, or modulus of elasticity, is the most important property describing the deformability of rock material. In this paper, grain-supported carbonates from Croatia are described and their elastic modulus and significant physical and mechanical properties are determined. The analysis of the collected data was performed in the R statistical environment. Estimation models based on multiple linear regression and the regression tree model were created. The methodology of model development and evaluation in R environment is described in detail. According to the more stringent coefficients (RMSECV and adjusted R2) used to evaluate the success of the estimation, simple regression tree models were found to perform well for the preliminary estimation, while more complex models based on Bagging performed very well. [ABSTRACT FROM AUTHOR]

Great progress has been made on machine learning and its applications are expanding rapidly nowadays. Through the case study of optimizing a magnetron injection gun for gyrotron devices, the functions of machine learning were investigated by using two supervised learning algorithms, regression trees and artificial neural networks. They showed excellent performance in predicting the outputs, exploring the importance of the input parameters and the relationship with the output parameters. Machine learning can be a useful tool in the development of microwave vacuum electron devices. [ABSTRACT FROM AUTHOR]

Cloud computing provides effective ways to rapidly provision computing resources over the Internet. For a better management of resource provisioning, the system requires to predict service-level agreements (SLAs) such as virtual machine (VM) startup times under various conditions of computing resources. The VM startup time is an important SLA parameter, which can impact other SLA parameters such as service initiation time and VM scale out times. By predicting VM startup times, Cloud providers can improve Cloud users' expectations. Various quality of service (QoS) parameters have been considered in different resource allocation frameworks. Also, there are several efforts addressing QoS prediction in cloud environments. However, little research has considered VM startup time as a QoS parameter. In this paper, we propose a regression tree model for predicting average, minimum, and maximum VM startup times. To test the efficiency of our model, we implemented the model in an OpenStack test environment. The test results show that our model predicts VM startup times with an average accuracy of 91.81%. [ABSTRACT FROM AUTHOR]

The structural failure induced by the localized corrosion of steel members frequently occurred. In this study, a spatial statistical technique was developed to predict the time-dependent localized corrosion initiated at the boundary zone of steel and concrete. 20 specimens were prepared by embedding steel plate into concrete block, the accelerated corrosion tests of 600, 1200, 1800, and 2400 cycles were conducted to obtain the actual surface figuration with uniform and localized corrosion. After performing the regression tree analysis to divide the corrosion surface into localized and uniformed corrosion regions, a semi-variogram was used as the statistical technique to generate the corrosion prediction model, and an ordinary Kriging model to estimate the probable corrosion depth at arbitrary locations. The simulated results show that the probability distribution of corrosion depth and surface topography of both localized and uniform corrosion surfaces are highly consistent with the test. Also, the estimated surfaces owned a similar stress concentration effect to the actual surface, although their localized corrosion distributes differently. The prediction results of 2400–9600 cycles show that the top five SCF values of simulation almost within the 95% confidence interval of the fitting curves of test results. Therefore, the accuracy of the estimated corrosion surface is considered reasonable using the spatial statistical simulation method. [ABSTRACT FROM AUTHOR]

Machine learning offers empirical methods to sift through accounting datasets with a large number of variables and limited a priori knowledge about functional forms. In this study, we show that these methods help detect and interpret patterns present in ongoing accounting misstatements. We use a wide set of variables from accounting, capital markets, governance, and auditing datasets to detect material misstatements. A primary insight of our analysis is that accounting variables, while they do not detect misstatements well on their own, become important with suitable interactions with audit and market variables. We also analyze differences between misstatements and irregularities, compare algorithms, examine one-year- and two-year-ahead predictions and interpret groups at greater risk of misstatements. [ABSTRACT FROM AUTHOR]

Short-term electric power load forecasting is a critical and essential task for utilities in the electric power industry for proper energy trading, which enables the independent system operator to operate the network without any technical and economical issues. From an electric power distribution system point of view, accurate load forecasting is essential for proper planning and operation. In order to build most robust machine learning model to forecast the load with a good accuracy irrespective of weather condition and type of day, features such as the season, temperature, humidity and day-status are incorporated into the data. In this paper, a machine learning model, namely a regression tree, is used to forecast the active power load an hour and one day ahead. Real-time active power load data to train and test the machine learning models are collected from a 33/11 kV substation located in Telangana State, India. Based on the simulation results, it is observed that the regression tree model is able to forecast the load with less error.

Questions: Inundation is an important feature linked to monodominance in tropical wetlands. The occurrence of neighbouring monodominant species indicates that soil can also have an important role. This study aimed to determine if variations in densities of three neighbouring monodominant species, Byrsonima cydoniifolia, Copernicia alba and Tabebuia aurea, are related primarily to variation of flooding patterns and secondarily to soil physicochemical variations in the Pantanal. Location: This study was performed in the southern Pantanal subregion of Miranda. Methods: We sampled 74 plots of 25 m × 25 m distributed among 15 patches of these species (five patches of each species). In each plot, we sampled all individuals of all shrub and tree species, checked the water mark left on tree trunks by the last flooding, and collected soil samples. Data were analyzed using canonical correspondence analysis (CCA) and regression trees. Results: We found that the main driver splitting these monodominant stands is soil, and that the secondary driver is inundation. The main drivers for Copernicia alba were higher soil pH, higher inundation and soil contents of Mg. For Byrsonima cydoniifolia, the main drivers were low Ca and Mg contents and high content of fine sand. Tabebuia aurea was the only species for which inundation was the main driver. Conclusions: We reject the hypothesis that inundation is the main driver splitting these monodominant stands. Differences in flooding levels were not crucial for two of the three species. Except for Tabebuia aurea, which was more influenced by inundation, soils were found to play an essential role in this spatial separation. [ABSTRACT FROM AUTHOR]

Predicting the long-term functional outcome after traumatic spinal cord injury (TSCI) is needed to adapt medical strategies and plan an optimized rehabilitation. This study investigates the use of regression trees for the development of predictive models based on acute clinical and demographic predictors. This prospective study was performed on 172 patients hospitalized after TSCI. Functional outcome was quantified using the Spinal Cord Independence Measure (SCIM) collected within the first-year post-injury. Age, delay before surgery, and Injury Severity Score (ISS) were considered as continuous predictors whereas energy of injury, trauma mechanisms, neurological level of injury, injury severity, occurrence of early spasticity, urinary tract infection, pressure ulcer, and pneumonia were coded as categorical inputs. A simplified model was built using only American Spinal Injury Association Impairment Scale grade, neurological level, energy, and age as predictor and was compared to a more complex model considering all 11 predictors mentioned above. The models built using 4 and 11 predictors were found to explain 51.4% and 62.3% of the variance of the SCIM total score after validation, respectively. Severity of the neurological deficit at admission was found to be the most important predictor. Other important predictors were the ISS, age, neurological level, and delay before surgery. Regression trees offer promising performances for predicting the functional outcome after a TSCI. It could help to determine the number and type of predictors leading to a prediction model of the functional outcome that can be used clinically in the future. [ABSTRACT FROM AUTHOR]

Regression trees are powerful tools in data mining for analyzing data sets. Observations are usually divided into homogeneous groups, and then statistical models for responses are derived in the terminal nodes. This paper proposes a new approach for regression trees that considers the dependency structures among covariates for splitting the observations. The mathematical properties of the proposed method are discussed in detail. To assess the accuracy of the proposed model, various criteria are defined. The performance of the new approach is assessed by conducting a Monte-Carlo simulation study. Two real data sets on classification and regression problems are analyzed by using the obtained results. [ABSTRACT FROM AUTHOR]

Random forest models have recently gained popularity for economic forecasting. Earlier studies demonstrated their potential to provide early warnings of recession and serve as a competitive method to older prediction models. This study offers the first evaluation of the random forest forecast for the Czech economy. The one-step-ahead forecasting results show high accuracy on the Czech data and are proven to outperform forecasts from the Czech Ministry of Finance and the Czech National Bank. The following multistep random forest forecast, estimated for the next four quarters, shows results similar to those from the central institutions. The main difference stems from the household and industrial confidence variables, which significantly impact on the random forest forecast. The variable-importance analysis further emphasizes the soft variables as valuable determinants for Czech forecasting. Overall, the findings motivate other forecasters to exercise this method. [ABSTRACT FROM AUTHOR]

The purpose of this study was to capture some factors affecting wool fineness (WF) in Karacabey Merino. For this goal, CHAID (Chi-Square Automatic Interaction Detector) tree-based algorithm implemented to construct a regression tree diagram was specified based on Bonferroni adjustment within the scope of the prediction of wool fineness as a response variable. Also, Multivariate Adaptive Regression Splines (MARS) was implemented for the WF prediction. In the prediction of wool fineness (WF), sex, dam age (DA), birth weight (BW), birth type (BT), live body weight (LBW), greasy fleece weight (GFW), staple length (SL), number of fibers (F) and average number of crimps over a length of 5 cm (ANC) were considered as possible predictors. To guarantee the highest predictive accuracy of the CHAID algorithm, minimum animal numbers in parent and child nodes were thought as 4 and 2. Model fit statistics showed the powerful predictive performance of the CHAID and MARS algorithms, but MARS outperformed CHAID. Considering the regression tree diagram generated by CHAID algorithm, the most influential predictor affecting WF was F, followed by BW, ANC and DA at the 2nd significance degree, and SL at the 3rd significance degree, respectively. MARS predictive model with the selected 5 terms captured only F as a significant predictor. In conclusion, CHAID and MARS data mining algorithms reflected that F predictor may be considered as an indirect selection criterion in the characterization of the breed standards of the Karacabey Merino in wool characteristics for breeding goals. [ABSTRACT FROM AUTHOR]

This article presents an exciting application of machine learning for loan origination in microfinance. Microfinance targets people who cannot build a credit history and therefore cannot access loans from banks or other financial institutions. We use data from a Mexican microfinance company that operates in several regions throughout the country. The objective is to guide intermediate lenders to choose their clients and achieve a lowerr credit default risk. We use several statistical models such as principal component analysis, clustering analysis and a regression tree. We obtain, as a result, a series of recommendations based on the characteristics of the clients. [ABSTRACT FROM AUTHOR]

We model the conditional risk premium by combining principal component analysis and a statistical learning technique, known as boosted regression trees. The method is validated through various out‐of‐sample tests. We apply the estimates to test the positivity restriction on the risk premium and find evidence that the risk premium is negative in periods of low corporate and government bond returns, high inflation and downward‐sloping term structure. These periods are linked with changes in business cycles; the states when theories predict the existence of negative risk premium. Based on the evidence, we reject the conditional capital asset pricing model and raise a question over the practice of imposing the positive risk premium constraint in predictive models. [ABSTRACT FROM AUTHOR]

Copyright of Vie et Sciences de l'Entreprise is the property of ANDESE and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

This study determined the association between Myf5, CAST and MSTN genes and fattening performance traits in 228 (Holstein = 103 and Brown = 125) cattle breeds. Classification and regression tree (CART) was used to determine association between genes and fattening performance. The allele frequencies of Holstein cattle in terms of Myf5, CAST and MSTN genes were A:0.30, B:0.70; A:0.43, G:0.56; A:0.97, B:0.03, respectively, whereas in the Brown Swiss cattle were A:0.36, B:0.64; A:0.65, G:0.34; A:0.88, B:0.12 respectively. The Myf5, CAST, and MSTN genes were found to be in the Hardy–Weinberg equilibrium (p > 0.05) in both of the cattle breeds. As a result, the association between Myf5, CAST and MSTN genes and fattening performance traits were found to be significant according to CART analysis. [ABSTRACT FROM AUTHOR]

Data Mining plays a decisive role especially in medical domain. Decision trees are predominant model in machine learning. Decision trees are simple and very effective classification approach. The decision tree identifies the utmost prime features of a given problem. One of the most common disease in India is Liver Cirrhosis. It is distinctly difficult to uncover Liver Cirrhosis in its initial stage. However early diagnosis of Liver Cirrhosis is highly important. The liver disease data set has a collection of distinguishing features that affect the healthy state of a patient. Machine Learning methods enable knowledge acquisition in early stages and use of this acquired knowledge plays an important role in solving problems like suppose if we want to predict whether the patient with the Liver Cirrhosis has also been suffering from Hepatitis C or not. In order to easily arrive at this knowledge certainly there is a need for fully integrated system. In this paper the collected Liver disease data set is analyzed and prognosticated whether the patient is suffering from liver cirrhosis or not. [ABSTRACT FROM AUTHOR]

Data-driven modeling of dynamical systems gathers attention in several applications; in conjunction with model predictive control, novel different identification techniques that merge machine learning and optimization are presented and compared with the purpose of reducing seismic response of frame structures and minimize control effort. Performance of neural network-, random forest- and regression tree-based identification algorithms in producing reliable models exploiting historical data coming from a real structure is shown. Peculiarities of each data-driven-based model emphasizing the strong potentialities of such approaches are highlighted, and it is shown in a simulative environment how, by slightly increasing the complexity of a model via random forests, we can reduce by half the active control effort with respect to the control computed exploiting regression trees-based models. [ABSTRACT FROM AUTHOR]

Typhoons have substantial impacts on power systems and may result in major power outages for distribution network users. Developing prediction models for the number of users going through typhoon power outages is a high priority to support restoration planning. This study proposes a data‐driven model to predict the number of distribution network users that may experience power outages when a typhoon passes by. To improve the accuracy of the prediction model, twenty six explanatory variables from meteorological factors, geographical factors and power grid factors are considered. In addition, the authors compared the application effect of five different machine learning regression algorithms, including linear regression, support vector regression, classification and regression tree, gradient boosting decision tree and random forest (RF). It turns out that the RF algorithm shows the best performance. The simulation indicates that the accuracy of the optimal model error within ±30% can reach up to 86%. The proposed method can improve the prediction accuracy through continuous learning on the existing basis. The prediction results can provide efficient guidance for emergency preparedness during typhoon disaster, and can be used as a basis to notify the distribution network users who are likely to lose power. [ABSTRACT FROM AUTHOR]

Coronavirus disease 2019 (COVID-19) has developed into a global pandemic, affecting every nation and territory in the world. Machine learning-based approaches are useful when trying to understand the complexity behind the spread of the disease and how to contain its spread effectively. The unsupervised learning method could be useful to evaluate the shortcomings of health facilities in areas of increased infection as well as what strategies are necessary to prevent disease spread within or outside of the country. To contribute toward the well-being of society, this paper focusses on the implementation of machine learning techniques for identifying common prevailing public health care facilities and concerns related to COVID-19 as well as attitudes to infection prevention strategies held by people from different countries concerning the current pandemic situation. Regression tree, random forest, cluster analysis and principal component machine learning techniques are used to analyze the global COVID-19 data of 133 countries obtained from the Worldometer website as of April 17, 2020. The analysis revealed that there are four major clusters among the countries. Eight countries having the highest cumulative infected cases and deaths, forming the first cluster. Seven countries, United States, Spain, Italy, France, Germany, United Kingdom, and Iran, play a vital role in explaining the 60% variation of the total variations by us of the first component characterized by all variables except for the rate variables. The remaining countries explain only 20% of the variation of the total variation by use of the second component characterized by only rate variables. Most strikingly, the analysis found that the variable number of tests by the country did not play a vital role in the prediction of the cumulative number of confirmed cases.

Acid phosphatase produced by plants and microbes plays a fundamental role in the recycling of soil phosphorus (P). A quantification of the spatial variation in potential acid phosphatase activity (AP) on large spatial scales and its drivers can help to reduce the uncertainty in our understanding of bio-availability of soil P. We applied two machine-learning methods (Random forests and back-propagation artificial networks) to simulate the spatial patterns of AP across Europe by scaling up 126 site observations of potential AP activity from field samples measured in the laboratory, using 12 environmental drivers as predictors. The back-propagation artificial network (BPN) method explained 58% of AP variability, more than the regression tree model (49%). In addition, BPN was able to identify the gradients in AP along three transects in Europe. Partial correlation analysis revealed that soil nutrients (total nitrogen, total P, and labile organic P) and climatic controls (annual precipitation, mean annual temperature, and temperature amplitude) were the dominant factors influencing AP variations in space. Higher AP occurred in regions with higher mean annual temperature, precipitation and higher soil total nitrogen. Soil TP and Po were non-monotonically correlated with modeled AP for Europe, indicating diffident strategies of P utilization by biomes in arid and humid area. This study helps to separate the influences of each factor on AP production and to reduce the uncertainty in estimating soil P availability. The BPN model trained with European data, however, could not produce a robust global map of AP due to the lack of representative measurements of AP for tropical regions. Filling this data gap will help us to understand the physiological basis of P-use strategies in natural soils.

Conservation agriculture, also known as environment-friendly agriculture, is expected to contribute to global climate change mitigation and biodiversity conservation. To understand the effect of conservation agriculture on farmers and identify those factors, such as farmers’ income change, that might affect practices of conservation agriculture, perceptions, and output, this study examined farmers’ economic and behavioral factors, motivation, and satisfaction. We surveyed 51 farmers who are receiving subsidies to practice conservation agriculture in Ishikawa Prefecture, Japan. The survey is one of the first prefectural-scale studies that combines unique quantitative analysis of motivation and satisfaction levels (e.g., behaviors) in temporal sequence from the initial to current time to practice conservation agriculture. Our results showed that years of experience, trade with a retail shop, and the farmer’s age can affect income change. With regard to social factors, the satisfaction of their fellowship with other farmers practicing conservation agriculture was also significantly correlated with income change. Simultaneously, this category of satisfaction was difficult to attain compared to the other categories. Thus, greater effort is needed to enhance support networking among conservation farmers. Furthermore, the work presented here also provides the opportunity for future research on temporal and spatial questions surveying economic and behavioral effects with consideration of the heightened policy promotion and entrance of large retail industries.

The Guiana dolphin (Sotalia guianensis) is a small-bodied dolphin distributed along the Atlantic coast from Honduras to southern Brazil. It preys on fish, squid and shrimps. Several seminal studies have described its diet, yet relationships between the species’ feeding plasticity and climate gradients remain unknown. We compiled a large database of Guiana dolphin stomach remains from southeast coastal Brazil. We described the species’ diet using a number of descriptors, multivariate analysis of variance to test possible differentiation in diet composition, and the Morisita index to estimate the extent of trophic niche overlap between groups. We also analysed feeding plasticity using a regression tree analysis followed by an ordination analysis. We present new records of prey for the species in Brazil. Our results suggest that the Guiana dolphin has opportunistic feeding habits, which may exhibit the species’ feeding plasticity. Such feeding plasticity is associated with the capability to prey throughout a wide array of climate conditions. From a conservation ecology perspective, we conclude that estuaries—even ones that are over-depleted and succumbing to human impacts—are paramount environments for the Guiana dolphin, serving as important sources of prey for the species and other sympatric marine mammals.

Abstract A debutanizer column is an integral part of any petroleum refinery. Online composition monitoring of debutanizer column outlet streams is highly desirable in order to maximize the production of liquefied petroleum gas. In this article, data-driven models for debutanizer column are developed for real-time composition monitoring. The dataset used has seven process variables as inputs and the output is the butane concentration in the debutanizer column bottom product. The input–output dataset is divided equally into a training (calibration) set and a validation (testing) set. The training set data were used to develop fuzzy inference, adaptive neuro fuzzy (ANFIS) and regression tree models for the debutanizer column. The accuracy of the developed models were evaluated by simulation of the models with the validation dataset. It is observed that the ANFIS model has better estimation accuracy than other models developed in this work and many data-driven models proposed so far in the literature for the debutanizer column.

Automatic gaze estimation not based on commercial and expensive eye tracking hardware solutions can enable several applications in the fields of human computer interaction (HCI) and human behavior analysis. It is therefore not surprising that several related techniques and methods have been investigated in recent years. However, very few camera-based systems proposed in the literature are both real-time and robust. In this work, we propose a real-time user-calibration-free gaze estimation system that does not need person-dependent calibration, can deal with illumination changes and head pose variations, and can work with a wide range of distances from the camera. Our solution is based on a 3-D appearance-based method that processes the images from a built-in laptop camera. Real-time performance is obtained by combining head pose information with geometrical eye features to train a machine learning algorithm. Our method has been validated on a data set of images of users in natural environments, and shows promising results. The possibility of a real-time implementation, combined with the good quality of gaze tracking, make this system suitable for various HCI applications.

Background & Objective: Migration, in any forms and by any motivations or outcomes, as a demographic phenomenon, has various cultural and socio-economic effects on local, regional, national and international levels. On the other hand, fertility plays an important role in health and population studies and researchers have examined its changes and trends in various aspects. The aim of this research was modeling the mean number of children ever born (CEB) for women who have left their cities or villages and migrated to Tehran city using regression tree model. Methods: Data was obtained from 2% of raw data from the census of 2011 and analyzed by regression tree model. Tree models are nonparametric statistical techniques which do not need complicated and unreachable assumptions of traditional parametric ones and have a considerable accuracy of modeling. These models are associated with simple interpretation of results. Therefore, they have been used by researches in many fields such as social sciences. Results: Age, educational level, job status, cause of migration, internet use for urban migrant women and age for rural migrant women were assumed as influential covariates in predicting the mean number of CEB. Conclusion: Regression tree findings revealed that urban migrants who were in higher age groups, lower educational levels, unemployed and have not used internet have had more mean number of CEBs.

The stability of wetlands is threatened by the combined effects of global climate change and human activity. In particular, the vegetation cover status of lake wetlands has changed. Here, the change in vegetation cover at the estuary of Poyang Lake was monitored, and its influencing factors are studied to elucidate the dynamic change characteristics of vegetation at the inlet of this lake. Flood and water level changes are two of the main factors affecting the evolution of wetland vegetation at the estuary of Poyang Lake. Therefore, Landsat data from 2000 to 2019 were used to study the spatial and temporal variation in the Normalized Difference Vegetation Index (NDVI) in the vegetation cover area. Theil–Sen Median trend analysis and Mann–Kendall tests were used to study the long-term trend characteristics of NDVI. The response between NDVI and the explanatory variables at the estuary of Poyang Lake was quantified using regression tree analysis to study the regional climate, water level, and flood inundation duration. Results showed the following: (1) Vegetation in a large area of the study area improved significantly from 2000 to 2010 and only slightly from 2010 to 2019, and few areas with slight degradation of vegetation were found. In most of these areas, the vegetation from 2000 to 2010 exhibited a gradual change, from nothing to something, which started around 2004; (2) The main variable that separated the NDVI values was the mean water level in October. When the mean October water level was greater than 14.467 m, the study area was still flooded in October. Thus, the regional value of BestNDVI was approximately 0.3, indicating poor vegetation growth. When the mean water level in October was less than 14.467 m, the elevation of the study area was higher than the water level value, and after the water receded in October, the wetland vegetation exhibited autumn growth in that year. Thus, the vegetation in the study area grew more abundantly. These results could help manage and protect the degraded wetland ecology.

Objective To construct the prediction index of anxiety outcome of nurses in operating room by using classified regression tree model. Methods A total of 284 operating room nurses in 971 th Hospital of Navy of Chinese People's Liberation Army from March 2017 to June 2020 were selected to evaluate the baseline set data by using the classification regression tree model to determine the risk factors inducing anxiety of operating room nurses in military hospitals, and calculate the contribution rate of different risk factors. The prediction set was used as the evaluation method of prediction results to determine the risk factors that induced nurses' self rating Anxiety Scale ( SAS * score to reach the anxiety standard. Results The results of single factor analysis confirmed that the weekly working time, marital status, having children, CD-RISC score, age educational background were statistically significant ( all P < 0. 05 *. The SAS score split regression tree analysis of nurses showed that the weekly working time, marital status, children's situation, CD-RISC score, age entry model, formed 6 nodes, among which there were divorce, weekly working time >45 h and the highest score. Conclusion The risk factors of nurses anxiety in operating room were composed of five items: working hours, marital status, children's condition, psychological elasticity and age. Nursing managers can use the risk factors of nurses anxiety in operating room to predict the anxiety outcome in advance and carry out timely and reasonable intervention, which can help to reduce the anxiety degree of nursing staff. [ABSTRACT FROM AUTHOR]

Accurately prediction of lake level fluctuations is essential for water resources planning and management. In the present study, the potential of a novel method, deep echo state network (Deep ESN), is investigated for monthly lake level prediction and its results are compared with three data-driven methods, artificial neural networks (ANNs), extreme learning machine (ELM), and regression tree (Reg. Tree). The methods are validated using root mean square errors (RMSE), determination coefficient (R2), and Nash-Sutcliffe efficiency (NSE) criteria. The investigated method (Deep ESN) outperforms the ELM, ANNs, and Reg. Tree by improving accuracies by 61–62–96%, 10–14–84%, and 8–23–80% in prediction 1 month, 2 months, and 3 months ahead lake level fluctuations in terms of RMSE criteria, respectively. Also, accuracy of ELM, ANNs, and Reg. Tree was significantly increased using Deep ESN model by 1.1–1.1–443%, 1.1–1.6–250%, and 1.6–6.5–184% in terms of NSE indicator for different lead-time horizons. Among the ELM, ANNs, and Reg. Tree, the third method provides the worst predictions while the first method performs superior to the second one in all tree time horizons. [ABSTRACT FROM AUTHOR]

In veneer-drying facilities, controllers face many challenges to maintain desired parameters in the final product based on customer's needs. The major challenge is setting process parameters to control the temperature and humidity within the various sections in the drying machine to obtain the desired properties of the final product. The regression tree approach can be used to simplify the complex relationship among process and product variables for identifying critical factors for drying veneer and achieving the desired range of veneer temperature. In this study, we investigated veneer-drying conditions and the short-term effect of climatic variables on veneer temperature. We have shown a three-step process to develop an optimal regression tree for veneer temperature. From the developed optimal tree, we are able to identify the most important threshold points of predictor space and adjustment for the climatic variables on the temperature of veneer sheets. The findings of this study were further investigated in an industrial setting and the desired veneer temperatures were attained for the final product. This application shows that we can follow the path of the optimal tree to pinpoint the most desired veneer temperature outcome. The developed optimal tree is relatively easy to use and interpret to estimate the average response of veneer temperature. [ABSTRACT FROM AUTHOR]

This study shows a mathematical and statistical analysis to generate models based on multiple linear regression (MLR) and regression trees (RT) that allow a reliable prediction of the Mass Yield (MY) and the Higher Heating Value (HHV) of the final solid product obtained by Hydrothermal Carbonization, called hydrochar. MLR models were obtained for lignocellulosic and non-lignocellulosic biomass using a set of experimental data with more than 500 points collected from the literature. A new approach based on dimensionless groups of variables that describe the composition of biomass and operational conditions was used. The analysis for each equation indicated that the MY depends on the process conditions and the biomass composition, which is proportional to the Polarity Index (IP) and Reactive Index (IR) values. On the other hand, the severity factor (log Ro) and the initial calorific value (HHVo) were the main factors for the HHV, but also the raw biomass composition (IP and H/C ratio) had an opposite and equal significant effect. For these equations, the results indicated an adjusted R2 (R2 a) of about 0.90 and an average RMSE of 6% and 1.7 MJ/kg for MY and HHV, respectively. Besides, explanatory variables were analyzed by their Relative Importance for the RT models. The severity factor (65%) and the IR (18%) were the most decisive variable in the MY prediction. The R2 and RMSE were 0.73 and 2%, respectively. For HHV, the variables with the most significant impact were the HHVo (33%), the log Ro (24%), and the IP (22%). In this case, the R2 and RMSE were 0.87 and 0.68 MJ/kg, respectively. Therefore, the model equations obtained are a powerful tool to predict the mass yield and the energetic value of the hydrochar before developing an experimental study. • Hydrochar mass yield (MY) depends on HTC process conditions and biomass reactivity. • HHV depends on the initial energy content of raw biomass and operational conditions. • Linear regression had R2 above 0.90 with RMSE of 6%. • Regression trees had R2 about 0.80 with RMSE of 2%. • Equations allow screening of process conditions before the experimental study. [ABSTRACT FROM AUTHOR]

Growth response of conifers is species- specific and depends on site and climate conditions. Studies on daily radial stem growth use different analytical approaches to determine species reactions to environmental conditions. These results contribute to improve forecasts of tree growth under a changing climate. During 2013 and 2014, radial stem growth of 33 mature Cedrus libani individuals growing under different climatic conditions in Turkey and Germany was monitored hourly using high precision point dendrometers. Stem radius increments (SRI) were extracted from dendrometer readings. The annual course of SRI showed site-specific patterns with mean daily values ranging between 9.9 and 29.3 µm over the growing season. Correlation and principal component analyses indicated that humidity and low temperatures during the growing season favored SRI. Multiple regression analyses demonstrated that precipitation and relative air humidity were the most important factors influencing daily SRI. Climate-growth relationships were further evaluated using the regression tree method. Precipitation was the most significant factor on daily SRI for all sites. The close coupling of SRI to relative air humidity and precipitation underlines the importance of stem water status for radial stem growth of C. libani which is native to regions with summer drought. It further explains the superior growth of C. libani in Germany. [ABSTRACT FROM AUTHOR]

Predictive modelling corroborates decision-making in the development of a standard habitat assessment protocol. In this study, we modelled environmental requirements of benthic macroinvertebrates. Classification and regression tree models (CART) and ordination analysis were performed to identify important variables affecting macroinvertebrate community pattern in the Lake Tana Watershed. A dataset of 95 samples was collected from eight wetlands. Among the modelled taxa, Coenagrionidae and Libellulidae had substantial predictive performance based on Kappa statistic (ĸ > 0.6) whereas Baetidae, Physidae, Tipulidae and Hydrophilidae had moderate predictive model performance (ĸ ≥ 0.4). Vegetation cover, leather tanning, vegetation clearance and nitrate ion were the topmost selected environmental variables influencing the occurrence of macroinvertebrate taxa. The conditional analysis depicted that the abundance of Coenagrionidae and Libellulidae increased with the increasing in vegetation cover. Overall, macroinvertebrate taxa have a clear habitat requirement within the habitat gradient studied and hence, could be a potential candidate for biomonitoring and provide valuable information in the development of a standard wetland assessment protocol. [ABSTRACT FROM AUTHOR]

The problem of dependency modeling by experimentally obtained observations is considered. The objective is to develop methods for neural network model synthesis allowing to automatize, simplify and speed-up model building. The mathematical support for neural network model synthesis is developed. It contains set of methods that transform the sample into a decision tree or a regression tree, on the basis of which the neural network structure is formed and the parameters are adjusted. The experiments on practical problems solving were carried out. Their results were confirmed the efficiency of the proposed methods. The results of the experiments allow to recommend the developed methods for solving the problems of constructing neural network models on precedents. [ABSTRACT FROM AUTHOR]

Dilated cardiomyopathy (DCM) is a heart muscle disease characterized by left ventricular (LV) or biventricular dilatation and systolic dysfunction in the absence of either pressure or volume overload or coronary artery disease sufficient to explain the dysfunction. The use of heart rate variability (HRV) analysis as well as of some machine learning algorithms, proved to be a valuable support in the diagnosis of cardiovascular disease. However, till now, only single beats or electrocardiogram segments of subjects affected by DCM were identified using machine learning techniques applied to HRV parameters. In this study, we used linear and non-linear HRV parameters and some clinical parameters (age, sex and left ventricular ejection fraction) evaluated on a large cohort of 972 subjects to early identify subjects suffered from DCM and to find which features could be selected as predictors for a correct diagnosis. By using principal component analysis and stepwise regression, we reduced the original parameters used as inputs for a series of classification and regression trees (CART). The highest accuracy of 97% and Area Under the Curve (AUC) of 95% were achieved using the ratio between low frequency and high frequency (LF/HF), sex and left ventricular ejection fraction (LVEF) parameters as inputs of the classifier. [ABSTRACT FROM AUTHOR]

Generalized linear models have the important property of providing unbiased estimates on a portfolio level. This implies that generalized linear models manage to provide accurate prices on a portfolio level. On the other hand, neural networks may provide very accurate prices on an individual policy level, but state-of-the-art use of neural networks does not pay any attention to unbiasedness on the portfolio level. This is an implicit consequence of applying early stopping rules in gradient descent methods for model fitting. In the present paper we discuss this deficiency and we provide two different techniques to overcome this drawback of neural network model fitting. [ABSTRACT FROM AUTHOR]

This article represents an attempt at spatiotemporal modelling of transport demand in a metropolitan area over the period of six years. We employed classical forms of machine learning and created regression models, which predict bike-sharing demand based on daily cumulative sums of bike trips for each considered docking station in the area. We next derived hourly estimates of demand by considering the statistical distribution of demand across individual hours of an average day. Longer-term distribution characteristics of bike trips, such as their average number falling on each day of the week, month of the year, etc., were also fed to these models as input attributes. We validated the models by Hold-Out as well as 10-Fold Cross-Validation, with promising results. [ABSTRACT FROM AUTHOR]

Summary: Graphics Processing Units (GPUs) are used as accelerators for improving performance while executing highly data parallel applications. The GPUs are characterized by a number of Streaming Multiprocessors (SM) and a large number of cores within each SM. In addition to this, a hierarchy of memories with different latencies and sizes is present in the GPUs. The program execution in GPUs is thus dependent on a number of parameter values, both at compile time and runtime. To obtain the optimal performance with these GPU resources, a large parameter space is to be explored, and this leads to a number of unproductive program executions. To alleviate this difficulty, machine learning–based autotuning systems are proposed to predict the right configuration using a limited set of compile‐time parameters. In this paper, we propose a two‐stage machine learning–based autotuning framework using an expanded set of attributes. The important parameters such as block size, occupancy, eligible warps, and execution time are predicted. The mean relative error in prediction of different parameters ranges from of 16% to 6.5%. Dimensionality reduction for the features set reduces the features by up to 50% with further increase in prediction accuracy. [ABSTRACT FROM AUTHOR]

Multiple imputation by chained equations (MICE) has emerged as a leading strategy for imputing missing epidemiological data due to its ease of implementation and ability to maintain unbiased effect estimates and valid inference. Within the MICE algorithm, imputation can be performed using a variety of parametric or nonparametric methods. Literature has suggested that nonparametric tree-based imputation methods outperform parametric methods in terms of bias and coverage when there are interactions or other nonlinear effects among the variables. However, these studies fail to provide a fair comparison as they do not follow the well-established recommendation that any effects in the final analysis model (including interactions) should be included in the parametric imputation model. We show via simulation that properly incorporating interactions in the parametric imputation model leads to much better performance. In fact, correctly specified parametric imputation and tree-based random forest imputation perform similarly when estimating the interaction effect. Parametric imputation leads to slightly higher coverage for the interaction effect, but it has wider confidence intervals than random forest imputation and requires correct specification of the imputation model. Epidemiologists should take care in specifying MICE imputation models, and this paper assists in that task by providing a fair comparison of parametric and tree-based imputation in MICE. [ABSTRACT FROM AUTHOR]

In this research we analyze a new approach for prediction of demand. In the studiedmarket of performing arts the observed demand is limited by capacity of the house. Then one needs to account for demand censorship to obtain unbiased estimates of demand function parameters. The presence of consumer segments with different purposes of going to the theater and willingness-to-pay for performance and ticket characteristics causes a heterogeneity in theater demand. We propose an estimator for prediction of demand that accounts for both demand censorship and preferences heterogeneity. The estimator is based on the idea of classification and regression trees and bagging prediction aggregation extended for prediction of censored data. Our algorithm predicts and combines predictions for both discrete and continuous parts of censored data. We show that our estimator performs better in terms of prediction accuracy compared with estimators which account either for censorship or heterogeneity only. The proposed approach is helpful for finding product segments and optimal price setting. [ABSTRACT FROM AUTHOR]

Determination of mining capital cost always is a challenging issue for mining engineer. Underestimating the capital cost may postpone the construction and production phases. In addition, overestimating may decrease value of the project. Currently available capital cost estimation models cannot predict mining capital cost in a reliable range of error. Therefore, in this paper, a model based on regression tree method is developed to estimate the mining capital cost. According to the obtained results, the capability of presented model to estimate the mining capital cost in a wide range of mining capacity is significant. [ABSTRACT FROM AUTHOR]

Understanding the human–water–food–climate nexus is central to achieving sustainable intensification (SI) in agriculture. This research uses a socio-hydrological approach to understand the underpinning for implementing SI in the Gezira Irrigation Scheme, Sudan, by integrating vegetation indices derived from remote sensing, ancillary, gridded soil and precipitation data, supplemented by interviews with 393 farmers. The productivity gap was estimated as the difference between the potential and actual productivities. Based upon data on farmers' socio-economic status and field practices, a regression tree model was built to determine the factors that control the sorghum yield. The model revealed that the financial status of farmers and access to water are the most influential factors on sorghum yield. A conceptual framework that elucidates SI and its bi-directional feedback to the environment, society and the economy is proposed. Implementing SI in the scheme has implications on water and food security in Sudan and beyond its borders. [ABSTRACT FROM AUTHOR]

Short-term traffic flow forecasting is crucial for proactive traffic management and control. One key issue associated with the task is how to properly define and capture the temporal patterns of traffic flow. A feasible solution is to design a multi-regime strategy. In this paper, an effective approach to forecasting short-term traffic flow based on multi-regime modeling and ensemble learning is presented. First, to properly capture the different patterns of traffic flow dynamics, a regime identification model based on probabilistic modeling was developed. Each identified regime represents a specific traffic phase, and was used as the representative feature for the forecasting modeling. Second, a forecasting model built on an ensemble learning strategy was developed, which integrates the forecasts of multiple regression trees. The traffic flow data over 5-min intervals collected from four I-80 freeway segments, in California, USA, was used to evaluate the proposed approach. The experimental results show that the identified regimes are able to well explain the different traffic phases, and play an important role in forecasting. Furthermore, the developed forecasting model outperformed four typical models in terms of root mean square error (RMSE) and mean absolute percentage error (MAPE) on three traffic flow measures. [ABSTRACT FROM AUTHOR]

The Atlantic Forest is the second most diverse forest system in South America, and only a fraction of its original distribution remains. In this study, we aim to use robust datasets of small mammals along the entire forest system to disentangle the main drivers for diversity along this gradient. More specifically, we aim to disentangle whether deforestation (recent), biogeographical variables, including 19 bioclimatic variables (historic), or historical trapping bias best describe patterns of taxonomic, functional, and phylogenetic diversities using small mammal assembles, from northeastern Brazil to eastern Paraguay. For that, we applied regression tree analyses to determine what environmental variables best describe each of the dimensions of diversity. Additionally, we implemented polynomial regression to test nonlinear relationships between biodiversity metrics and patch size. We found that patterns of overall taxonomic, functional, and phylogenetic diversities; rodent taxonomic diversity; and marsupial functional diversity were better explained by temperature variables. Meanwhile, marsupial taxonomic and phylogenetic diversities, and rodent functional and phylogenetic diversities were best explained by precipitation variables. Furthermore, patch area, trapping, and latitude were never the best descriptors for any of the diversity dimensions. Although all dimensions of biodiversity are correlated, they have unique information and should be considered individually to better understand biodiversity and inform conservation strategies. We found that fragmentation is impactful at a local scale and becomes less important at a biogeographical scale. Therefore, climatic variables drove biogeographical faunal patterns for all clades, probably reflecting important historical assembly process at large spatial scales. [ABSTRACT FROM AUTHOR]