Best Model In Terms Research Articles

A novel method for fuzzy diagnostic system design.

Background: In recent years, liver disorders have been continuously increased. Proper performance of data mining techniques in decision-making and forecasting caused to use them commonly in designing of automatic medical diagnostic systems. The main aim of this paper is to introduce a classifier for diagnosis of liver disease that not only has high precision but also is understandable and has been created without expert knowledge. Methods: In regards to this purpose, fuzzy association rules have been extracted from dataset according to fuzzy membership functions which determined by fuzzy C-means clustering method; while each time, extracting fuzzy association rules, one of the five quality measures including confidence, coverage, reliability, comprehensibility and interestingness is used and five fuzzy rule-bases extracted based on them. Then, five fuzzy inference systems are designed on the basis of obtained rule-bases and evaluated in order to choose the best model in terms of diagnostic accuracy. Results: The proposed diagnostic method was examined using data set of Indian liver patients available at UCI repository. Results showed that among considered quality measures, interestingness, reliability and truth outperformed respectively, and yielded precision, sensitivity, specificity and accuracy of more than 90%. Conclusion: In this paper, a classification method was developed to predict liver disease which in addition to high classification accuracy, it has been created without expert knowledge and provided an understandable explanation of data. This method is convenient, user friendly, efficient and requires no expertise.

A comparative study of logistic regression based machine learning techniques for prediction of early virological suppression in antiretroviral initiating HIV patients

BackgroundTreatment with effective antiretroviral therapy (ART) lowers morbidity and mortality among HIV positive individuals. Effective highly active antiretroviral therapy (HAART) should lead to undetectable viral load within 6 months of initiation of therapy. Failure to achieve and maintain viral suppression may lead to development of resistance and increase the risk of viral transmission. In this paper three logistic regression based machine learning approaches are developed to predict early virological outcomes using easily measurable baseline demographic and clinical variables (age, body weight, sex, TB disease status, ART regimen, viral load, CD4 count). The predictive performance and generalizability of the approaches are compared.MethodsThe multitask temporal logistic regression (MTLR), patient specific survival prediction (PSSP) and simple logistic regression (SLR) models were developed and validated using the IDI research cohort data and predictive performance tested on an external dataset from the EFV cohort. The model calibration and discrimination plots, discriminatory measures (AUROC, F1) and overall predictive performance (brier score) were assessed.ResultsThe MTLR model outperformed the PSSP and SLR models in terms of goodness of fit (RMSE = 0.053, 0.1, and 0.14 respectively), discrimination (AUROC = 0.92, 0.75 and 0.53 respectively) and general predictive performance (Brier score= 0.08, 0.19, 0.11 respectively). The predictive importance of variables varied with time after initiation of ART. The final MTLR model accurately (accuracy = 92.9%) predicted outcomes in the external (EFV cohort) dataset with satisfactory discrimination (0.878) and a low (6.9%) false positive rate.ConclusionMultitask Logistic regression based models are capable of accurately predicting early virological suppression using readily available baseline demographic and clinical variables and could be used to derive a risk score for use in resource limited settings.

Stock Market Prediction Using Nonparametric Fuzzy and Parametric GARCH Methods

Prediction of stock market value is one the most complicated issue during the past decades. Due to its importance, in this research, we consider the prediction of stock values based on non-parametric and parametric methods. In this first method, we use the fuzzy Markov chain procedure in order to prediction problem. In this regard, all of the rising and falling probabilities during the weekdays are calculated and then they applied to obtain the increasing and decreasing rate. Then, based on this information we model and predict the stock values. In the sequel, we implement different methods of parametric time series such as generalized autoregressive conditionally heteroskedastic (GARCH), ARIMA-GARCH, Exponential GARCH (E-GARCH) and GJR-GARCH by assuming the normal and t-student distribution for the error terms to obtain the best model in terms of minimum mean square errors. Finally, the mythologies developed here are applied for the Tehran Stock Exchange Index (TEDPIX).

A case study on the use of machine learning techniques for supporting technology watch

Technology Watch human agents have to read many documents in order to manually categorize and dispatch them to the correct expert, that will later add valued information to each document. In this two step process, the first one, the categorization of documents, is time consuming and relies on the knowledge of a human categorizer agent. It does not add direct valued information to the process that will be provided in the second step, when the document is revised by the correct expert.This paper proposes Machine Learning tools and techniques to learn from the manually pre-categorized data to automatically classify new content. For this work a real industrial context was considered. Text from original documents, text from added value information and Semantic Annotations of those texts were used to generate different models, considering manually pre-established categories. Moreover, three algorithms from different approaches were used to generate the models. Finally, the results obtained were compared to select the best model in terms of accuracy and also on the reduction of the amount of document readings (human workload).

The applicability of credit scoring models in emerging economies: an evidence from Jordan

PurposeThe purpose of this study is to propose an objective and efficient method for assessing credit risk by introducing and investigating to a greater extent the applicability of credit scoring models in the Jordanian banks and to what range they can be used to achieve their strategic and business objectives.Design/methodology/approachThe research methodology comprises two phases. The first phase is the model development. Three modeling techniques are used to build the scoring models, namely, logistic regression (LR), artificial neural network (NN) and support vector machine (SVM), and the best performing model is selected for next stage. The second phase is two-fold: linking the credit expert knowledge in a way that can enhance the outcomes of the scoring model and a profitability test to explore if the selected model is efficient in meeting banks’ strategic and business objectives.FindingsThe findings showed that LR model outperformed both ANN and SVM across various performance indicators. The LR model also fits best with achieving the bank’s strategic and business objectives.Originality/valueTo the best of the authors’ knowledge, this study is the first that applied several modeling and classification techniques for Jordanian banks and calibrated the best model in terms of its strategic and business objectives. Furthermore, credit experts’ knowledge was engaged with the scoring model to determine its efficiency and reliability against the sole use of an automated scoring model in the hope to encourage the application of credit scoring models as an advisory tool for credit decisions.

Face detection using deep learning: An improved faster RCNN approach

In this report, we present a new face detection scheme using deep learning and achieve the state-of-the-art detection performance on the well-known FDDB face detetion benchmark evaluation. In particular, we improve the state-of-the-art faster RCNN framework by combining a number of strategies, including feature concatenation, hard negative mining, multi-scale training, model pretraining, and proper calibration of key parameters. As a consequence, the proposed scheme obtained the state-of-the-art face detection performance, making it the best model in terms of ROC curves among all the published methods on the FDDB benchmark.

Toward a spatial understanding of active transportation potential among a university population

ABSTRACTUniversities and surrounding communities stand to benefit when active travel mode choices are elevated. Despite this, there is little research on travel mode choice at commuter universities and, in particular, the nonlinear spatial relationships among active travel potential and various contextual and compositional factors. The purpose of this study was to examine and visualize linkages among personal, household, density, diversity, and design factors, and active travel (bicycling, walking, and mass-transit modes) among a commuter-university population residing throughout southeastern Michigan, USA. This was accomplished by employing exploratory spatial data analysis (ESDA), ordinary least squares (OLS) regression, and a geographically weighted regression (GWR) model. The GWR model outperformed the traditional OLS model in terms of goodness of fit (R2 = .534 and R2 = .461, respectively). A novel cartographic mapping technique was employed to depict where statistically significant parameter estimates negatively or positively influenced active travel. The main finding was that personal, household, density, diversity, and design estimates varied in both magnitude and spatiality throughout the university's study area. Interestingly, distance was not a universal barrier to active travel potential. These variations emphasize the importance of promoting active transportation through localized interventions as well as coordinating efforts among universities and surrounding communities.

Development of Mobile-Interfaced Machine Learning-Based Predictive Models for Improving Students’ Performance in Programming Courses

Student performance modelling (SPM) is a critical step to assessing and improving students performances in their learning discourse. However, most existing SPM are based on statistical approaches, which on one hand are based on probability, depicting that results are based on estimation; and on the other hand, actual influences of hidden factors that are peculiar to students, lecturers, learning environment and the family, together with their overall effect on student performance have not been exhaustively investigated. In this paper, Student Performance Models (SPM) for improving students performance in programming courses were developed using M5P Decision Tree (MDT) and Linear Regression Classifier (LRC). The data used was gathered using a structured questionnaire from 295 students in 200 and 300 levels of study who offered Web programming, C or JAVA at Federal University, Oye-Ekiti, Nigeria between 2012 and 2016. Hidden factors that are significant to students performance in programming were identified. The relevant data gathered, normalized, coded and prepared as variable and factor datasets, and fed into the MDT algorithm and LRC to develop the predictive models. The evaluation results obtained indicate that the variable-based LRC produced the best model in terms of MAE, RMSE, RAE and the RRSE having yielded the least values in all the evaluations conducted. Further results obtained established the strong significance of attitude of students and lecturers, fearful perception of students, erratic power supply, university facilities, student health and students attendance to the performance of students in programming courses. The variable-based LRC model presented in this paper could provide baseline information about students performance thereby offering better decision making towards improving teaching/learning outcomes in programming courses.

Generalized bent-cable methodology for changepoint data: a Bayesian approach

ABSTRACTThe choice of the model framework in a regression setting depends on the nature of the data. The focus of this study is on changepoint data, exhibiting three phases: incoming and outgoing, both of which are linear, joined by a curved transition. Bent-cable regression is an appealing statistical tool to characterize such trajectories, quantifying the nature of the transition between the two linear phases by modeling the transition as a quadratic phase with unknown width. We demonstrate that a quadratic function may not be appropriate to adequately describe many changepoint data. We then propose a generalization of the bent-cable model by relaxing the assumption of the quadratic bend. The properties of the generalized model are discussed and a Bayesian approach for inference is proposed. The generalized model is demonstrated with applications to three data sets taken from environmental science and economics. We also consider a comparison among the quadratic bent-cable, generalized bent-cable and piecewise linear models in terms of goodness of fit in analyzing both real-world and simulated data. This study suggests that the proposed generalization of the bent-cable model can be valuable in adequately describing changepoint data that exhibit either an abrupt or gradual transition over time.

FEASIBILITY OF MACHINE LEARNING METHODS FOR SEPARATING WOOD AND LEAF POINTS FROM TERRESTRIAL LASER SCANNING DATA

Abstract. Classification of wood and leaf components of trees is an essential prerequisite for deriving vital tree attributes, such as wood mass, leaf area index (LAI) and woody-to-total area. Laser scanning emerges to be a promising solution for such a request. Intensity based approaches are widely proposed, as different components of a tree can feature discriminatory optical properties at the operating wavelengths of a sensor system. For geometry based methods, machine learning algorithms are often used to separate wood and leaf points, by providing proper training samples. However, it remains unclear how the chosen machine learning classifier and features used would influence classification results. To this purpose, we compare four popular machine learning classifiers, namely Support Vector Machine (SVM), Na¨ıve Bayes (NB), Random Forest (RF), and Gaussian Mixture Model (GMM), for separating wood and leaf points from terrestrial laser scanning (TLS) data. Two trees, an Erytrophleum fordii and a Betula pendula (silver birch) are used to test the impacts from classifier, feature set, and training samples. Our results showed that RF is the best model in terms of accuracy, and local density related features are important. Experimental results confirmed the feasibility of machine learning algorithms for the reliable classification of wood and leaf points. It is also noted that our studies are based on isolated trees. Further tests should be performed on more tree species and data from more complex environments.

Biomarker comparison and selection for prostate cancer detection in Dynamic Contrast Enhanced-Magnetic Resonance Imaging (DCE-MRI)

In this work, the capability of imaging biomarkers obtained from multivariate curve resolution-alternating least squares (MCR-ALS), in combination with those obtained from first and second-generation pharmacokinetic models, have been studied for improving prostate cancer tumor depiction using partial least squares-discriminant analysis (PLS-DA). The main goal of this work is to improve tissue classification properties selecting the best biomarkers in terms of prediction. A wrapped double cross-validation method has been applied for the variable selection process. Using the best PLS-DA model, prostate tissues can be classified obtaining 13.4% of false negatives and 7.4% of false positives. Using MCR-ALS biomarkers yields the best models in terms of parsimony and classification performance.

Statistical Comparison of Growth Curve Models in First Years of Life in USA and Turkish Pediatric Populations

Background: In pediatric clinical trials and cohort studies, actual height and weight of children at a specific age may be required for certain developmental assessments such as energy expenditure. This necessitates the choice of a growth model with desired characteristics to predict height and weight accurately. Methods: We compared two commonly used growth curve models, namely, Logistic and Gompertz models, with respect to the distribution of their residuals as well as the logistical challenges in model convergence using the US and Turkish Growth Curve Standards for the first 3 years of life. We compared the model results in terms of the size of the residuals as well as prediction standard error for each data source, for each anthropometric measurement, namely, height and weight, for each gender, and for each modeling parameterizations. We also compared these models under missingness. Results: We have shown that Gompertz model with only the first parameter is defined with a profile specific random term as well is the best model in terms of prediction accuracy. Although the same Gompertz model fitted on each individual profile without a random term also has similar prediction accuracy, it has much inflated standard error of estimation, thus, not recommended to be used. Conclusion: We conclude that Gompertz model with only the first parameter is defined with as a random effect performs the best with and without missing data for both height and weight growth in the first three years of life.

The influence of non-stationarity in extreme hydrological events on flood frequency estimation

AbstractSubstantial evidence shows that the frequency of hydrological extremes has been changing and is likely to continue to change in the near future. Non-stationary models for flood frequency analyses are one method of accounting for these changes in estimating design values. The objective of the present study is to compare four models in terms of goodness of fit, their uncertainties, the parameter estimation methods and the implications for estimating flood quantiles. Stationary and non-stationary models using the GEV distribution were considered, with parameters dependent on time and on annual precipitation. Furthermore, in order to study the influence of the parameter estimation approach on the results, the maximum likelihood (MLE) and Bayesian Monte Carlo Markov chain (MCMC) methods were compared. The methods were tested for two gauging stations in Slovenia that exhibit significantly increasing trends in annual maximum (AM) discharge series. The comparison of the models suggests that the stationary model tends to underestimate flood quantiles relative to the non-stationary models in recent years. The model with annual precipitation as a covariate exhibits the best goodness-of-fit performance. For a 10% increase in annual precipitation, the 10-year flood increases by 8%. Use of the model for design purposes requires scenarios of future annual precipitation. It is argued that these may be obtained more reliably than scenarios of extreme event precipitation which makes the proposed model more practically useful than alternative models.

Analysis of intensity and spatial patterns of public use in natural treatment systems using geotagged photos from social media

Patterns of public use in 273 natural treatment systems worldwide are investigated by means of geotagged data from two popular photo-sharing websites, using spatial analysis and regression techniques. Standardized Major Axis (SMA) regression is found to perform better than other univariate calibration models in terms of goodness of fit with reported visitation frequencies and predictive accuracy, and is used to predict visitation rates in 139 systems that are associated with at least one geotagged photograph. High visitation rates are found in free-water surface (FWS) constructed wetlands and mixed pond-constructed wetlands systems, as well as systems treating surface water or stormwater runoff. Geographic Information System (GIS) techniques are used to map hot and cold spots of public use in two highly visited systems. Binomial logit regression reveals that the probability to be associated with at least one geotagged photograph is a function of system size, system type, and influent water quality. The findings are discussed in terms of their implications for the evaluation of public use in multifunctional ecologically engineered systems as well as the applicability of the proposed methodology to other natural and man-made ecosystems.

Multicategory Purchase Incidence Models for Partitions of Product Categories

We analyze multicategory purchases of households by means of heterogeneous multivariate probit models that relate to partitions formed from a total of 25 product categories. We investigate both prior and post hoc partitions. We search model structures by a stochastic algorithm and estimate models by Markov chain Monte Carlo simulation. The best model in terms of cross‐validated log‐likelihood refers to a post hoc partition with two groups; the second‐best model considers all categories as one group. Among prior partitions with at least two category groups a five‐group model performs best. Effects on average basket value differ for the model with five prior category groups from those for the best‐performing model in 40% and 24% of the investigated categories for features and displays, respectively. In addition, the model with five prior category groups also underestimates total sales revenue across all categories by about 28%. Copyright © 2016 John Wiley & Sons, Ltd.

Hierarchical models for the spatial–temporal carbon nanotube height variations

Carbon nanotubes (CNTs) are allotropes of carbon with a cylindrical nanostructure. Due to their low production cost and potentially high demand, the large-scale production of CNTs is urgently needed and will be highly profitable. However, quality control will be a great challenge to a large-scale production due to the delicate nature of the production process. Among the problems involved in the quality control of CNT array production, height variation is one of the primary concerns. The objective of this study is to model the height along both the spatial and temporal dimensions, so that the height variations can be controlled during the production process, thus improving the quality and stability of CNT arrays. Specifically, the height variation of the CNT arrays is decomposed into macro-scale and micro-scale variations. The macro-scale variation is modelled by state-space and regression models, and the micro-scale variation is modelled as a spatial process. The models successfully capture both the macro-trends and the micro-patterns. A practical case study shows the effectiveness of the proposed models in terms of goodness of fit and prediction accuracy, and the distinction of the models is summarised to aid in choosing a model to apply to other spatial–temporal data modelling problems.

Divisive heuristic for modularity density maximization

In this paper we consider a particular method of clustering for graphs, namely the modularity density maximization. We propose a hierarchical divisive heuristic which works by splitting recursively a cluster into two new clusters by maximizing the modularity density, and we derive four reformulations for the mathematical programming model used to obtain the optimal splitting. We report computational results of the eight algorithms (four reformulations with two different symmetry breaking strategies) obtained on some instances from the literature. Statistical tests show that the best model in terms of computational time is the one that is obtained with a dual reformulation of the bilinear terms arising in the objective function. Moreover, the hierarchical divisive heuristic provides generally near-optimal solutions in terms of modularity density.

Baking-drying kinetics of crisp bread: The influence of bran content and baking temperature

In this study, mathematical modeling of hot air baking-drying of thin-layer crisp bread was investigated. Thin-layer drying process were conducted under three different temperatures of 110, 150 and 190 °C at a constant air velocity of 0.5±0.1 m/s and absolute humidity of 0.6 ± 0.04g water/kg dry air. It was found that the baking-drying process occurred in falling rate period over the baking-drying times. Eight well-known thin-layer baking-drying models were fitted to the baking-drying experimental data of crisp bread, implementing non-linear regression analysis techniques. Based on the coefficient of determination (R²) and root mean square error (RMSE) values, it was concluded that the best models in terms of fitting performance for hot air baking-drying of bran free crisp bread were Wang & Singh and Logarithmic while for whole-wheat crisp bread were Page, Logarithmic and Wang & Singh. The moisture transfer from crisp bread was described using the Fick’s diffusion model. The effective diffusivity was within the range of 2.88×10-8 to 1.11×10-7 m2/s for bran free crisp bread and from 2.47×10-8 to 8.84×10-8 m2/s for whole-wheat crisp bread over the temperature range. The activation energy for bran free and whole-wheat crisp bread was found to be 25.22 and 23.43 kJ/mol, respectively..

Prediction of Core Body Temperature from Multiple Variables.

This paper aims to improve the prediction of rectal temperature (T re) from insulated skin temperature (T is) and micro-climate temperature (T mc) previously reported (Richmond et al., Insulated skin temperature as a measure of core body temperature for individuals wearing CBRN protective clothing. Physiol Meas 2013; 34:1531-43.) using additional physiological and/or environmental variables, under several clothing and climatic conditions. Twelve male (25.8±5.1 years; 73.6±11.5kg; 178±6cm) and nine female (24.2±5.1 years; 62.4±11.5kg; 169±3cm) volunteers completed six trials, each consisting of two 40-min periods of treadmill walking separated by a 20-min rest, wearing permeable or impermeable clothing, under neutral (25°C, 50%), moderate (35°C, 35%), and hot (40°C, 25%) conditions, with and without solar radiation (600W m(-2)). Participants were measured for heart rate (HR) (Polar, Finland), skin temperature (T s) at 11 sites, T is (Grant, Cambridge, UK), and breathing rate (f) (Hidalgo, Cambridge, UK). T mc and relative humidity were measured within the clothing. T re was monitored as the 'gold standard' measure of T c for industrial or military applications using a 10cm flexible probe (Grant, Cambridge, UK). A stepwise multiple regression analysis was run to determine which of 30 variables (T is, T s at 11 sites, HR, f, T mc, temperature, and humidity inside the clothing front and back, body mass, age, body fat, sex, clothing, Thermal comfort, sensation and perception, and sweat rate) were the strongest on which to base the model. Using a bootstrap methodology to develop the equation, the best model in terms of practicality and validity included T is, T mc, HR, and 'work' (0 = rest; 1 = exercise), predicting T re with a standard error of the estimate of 0.27°C and adjusted r (2) of 0.86. The sensitivity and specificity for predicting individuals who reached 39°C was 97 and 85%, respectively. Insulated skin temperature was the most important individual parameter for the prediction of T re. This paper provides novel information about the viability of predicting T c under a wide range of conditions, using predictors which can practically be measured in a field environment.

Is Not Doing the Washing Up Like Draft Dodging? The Military Model for Resisting a Gender Based Labour Division

AbstractI will examine a version of Bubeck's and Robeyns' proposals for ‘care duty’ which looks at the ways in which care work is analogous to defence work, and what the implications are for the best models in terms both of distributive justice and serving the common good. My own analysis will differ from Bubeck's and Robeyns' in two respects. First I will apply their arguments to all aspects of care including housework. This will mean making a case for housework counting as a form of care work as it is not usually regarded as such, and in particular, would probably be excluded from Robeyns' own account as she follows Bubeck's earlier characterisation of care as involving face‐to‐face interaction. Secondly, I will explore various ways in which care, and especially housework, could and has been distributed by appealing to a number of military models, concluding that the best gender‐just distribution of care requires a style of care experience modelled on universal military service. I will consider a number of objections to my view before concluding that not doing one's share of the housework could indeed come to be regarded as not doing one's duty as a citizen.