Class Of Regression Models Research Articles

A skew‐t quantile regression for censored and missing data

Quantile regression has emerged as an important analytical alternative to the classical mean regression model. However, the analysis could be complicated by the presence of censored measurements due to a detection limit of equipment in combination with unavoidable missing values arising when, for instance, a researcher is simply unable to collect an observation. Another complication arises when measures depart significantly from normality, for instance, in the presence of skew heavy‐tailed observations. For such data structures, we propose a robust quantile regression for censored and/or missing responses based on the skew‐t distribution. A computationally feasible EM‐based procedure is developed to carry out the maximum likelihood estimation within such a general framework. Moreover, the asymptotic standard errors of the model parameters are explicitly obtained via the information‐based method. We illustrate our methodology by using simulated data and two real data sets.

Improved method of estimating temperatures at meteor peak heights

Abstract. For 2 decades, meteor radars have been routinely used to monitor atmospheric temperature around 90 km altitude. A common method, based on a temperature gradient model, is to use the height dependence of meteor decay time to obtain a height-averaged temperature in the peak meteor region. Traditionally this is done by fitting a linear regression model in the scattered plot of log⁡10(1/τ) and height, where τ is the half-amplitude decay time of the received signal. However, this method was found to be consistently biasing the slope estimate. The consequence of such a bias is that it produces a systematic offset in the estimated temperature, thus requiring calibration with other co-located measurements. The main reason for such a biasing effect is thought to be due to the failure of the classical regression model to take into account the measurement error in τ and the observed height. This is further complicated by the presence of various geophysical effects in the data, as well as observational limitation in the measuring instruments. To incorporate various error terms in the statistical model, an appropriate regression analysis for these data is the errors-in-variables model. An initial estimate of the slope parameter is obtained by assuming symmetric error variances in normalised height and log⁡10(1/τ). This solution is found to be a good prior estimate for the core of this bivariate distribution. Further improvement is achieved by defining density contours of this bivariate distribution and restricting the data selection process within higher contour levels. With this solution, meteor radar temperatures can be obtained independently without needing any external calibration procedure. When compared with co-located lidar measurements, the systematic offset in the estimated temperature is shown to have reduced to 5 % or better on average.

Application of Spatial Modeling of Biophysical Variables in an Urbanized Area in the Amazon: The Case of the Metropolitan Area of Belém-Pará

Abstract The type of land use and land cover plays a decisive role in land surface temperature (LST). As cities are composed of varied covers, including vegetation, built-up areas, buildings, roads and areas without vegetation, understanding LST patterns in complex urban spaces is becoming increasingly important. The present study investigated the relationship between LST and albedo, NDVI, NDWI, NDBI and NDBaI in the period between 1994 and 2017. Images from Thematic Mapper (TM) and Thermal Infrared Sensor (TIRS) onboard the Landsat 5 and 8 satellites, respectively, were used in the study. The images were processed, resampled (spatial resolution of 120 m) in the environment of the QGIS 3.0 software and, finally, centroids were extracted resulting in a total of 1252 points. A classical regression (CR) model was applied to the variables, followed by spatial autoregressive (SARM) and spatial error (SEM) models, and the results were compared using accuracy indices. The results showed that the highest correlation coefficient was found between albedo and NDBaI (r = 0.88). The relationship between albedo and LST (r = 0.7) was also positive and significant at р < 0.05. The global Moran's I index showed spatial dependence and non-stationarity of the LST (I = 0.44). The SEM presented the best accuracy metrics (AIC = 3307.15 and R2 = 0.65) for the metropolitan region of Belém, explaining considerably more variations in the relationship between explanatory factors and LST when compared to conventional CR models.

Gingival Crevicular Placental Alkaline Phosphatase Is an Early Pregnancy Biomarker for Pre-Eclampsia.

Early and innovative diagnostic strategies are required to predict the risk of developing pre-eclampsia (PE). The purpose of this study was to evaluate the performance of gingival crevicular fluid (GCF) placental alkaline phosphatase (PLAP) concentrations to correctly classify women at risk of PE. A prospectively collected, retrospectively stratified cohort study was conducted, with 412 pregnant women recruited at 11–14 weeks of gestation. Physical, obstetrical, and periodontal data were recorded. GCF and blood samples were collected for PLAP determination by ELISA assay. A multiple logistic regression classification model was developed, and the classification efficiency of the model was established. Within the study cohort, 4.3% of pregnancies developed PE. GCF-PLAP concentration was 3- to 6-fold higher than in plasma samples. GCF-PLAP concentrations and systolic blood pressure were greater in women who developed PE (p = 0.015 and p < 0.001, respectively). The performance of the multiparametric model that combines GCF-PLAP concentration and the levels of systolic blood pressure (at 11–14 weeks gestation) showed an association of systolic blood pressure and GCF-PLAP concentrations with the likelihood of developing PE (OR:1.07; 95% CI 1.01–1.11; p = 0.004 and OR:1.008, 95% CI 1.000–1.015; p = 0.034, respectively). The model had a sensitivity of 83%, a specificity of 72%, and positive and negative predictive values of 12% and 99%, respectively. The area under the receiver operating characteristic (AUC-ROC) curve was 0.77 and correctly classified 72% of PE pregnancies. In conclusion, the multivariate classification model developed may be of utility as an aid in identifying pre-symptomatic women who subsequently develop PE.

Quantitative Detection of Chromium Pollution in Biochar Based on Matrix Effect Classification Regression Model.

Returning biochar to farmland has become one of the nationally promoted technologies for soil remediation and improvement in China. Rapid detection of heavy metals in biochar derived from varied materials can provide a guarantee for contaminated soil, avoiding secondary pollution. This work aims first to apply laser-induced breakdown spectroscopy (LIBS) for the quantitative detection of Cr in biochar. Learning from the principles of traditional matrix effect correction methods, calibration samples were divided into 1–3 classifications by an unsupervised hierarchical clustering method based on the main elemental LIBS data in biochar. The prediction samples were then divided into diverse classifications of calibration samples by a supervised K-nearest neighbor (KNN) algorithm. By comparing the effects of multiple partial least squares regression (PLSR) models, the results show that larger numbered classifications have a lower averaged relative standard deviations of cross-validation (ARSDCV) value, signifying a better calibration performance. Therefore, the 3 classification regression model was employed in this study, which had a better prediction performance with a lower averaged relative standard deviations of prediction (ARSDP) value of 8.13%, in comparison with our previous research and related literature results. The LIBS technology combined with matrix effect classification regression model can weaken the influence of the complex matrix effect of biochar and achieve accurate quantification of contaminated metal Cr in biochar.

Inverted Weibull Regression Models and Their Applications

In this paper, we propose the classical and Bayesian regression models for use in conjunction with the inverted Weibull (IW) distribution; there are the inverted Weibull Regression model (IW-Reg) and inverted Weibull Bayesian regression model (IW-BReg). In the proposed models, we suggest the logarithm and identity link functions, while in the Bayesian approach, we use a gamma prior and two loss functions, namely zero-one and modified general entropy (MGE) loss functions. To deal with the outliers in the proposed models, we apply Huber and Tukey’s bisquare (biweight) functions. In addition, we use the iteratively reweighted least squares (IRLS) algorithm to estimate Bayesian regression coefficients. Further, we compare IW-Reg and IW-BReg using some performance criteria, such as Akaike’s information criterion (AIC), deviance (D), and mean squared error (MSE). Finally, we apply the some real datasets collected from Saudi Arabia with the corresponding explanatory variables to the theoretical findings. The Bayesian approach shows better performance compare to the classical approach in terms of the considered performance criteria.

Boosting Poisson regression models with telematics car driving data

With the emergence of telematics car driving data, insurance companies have started to boost classical actuarial regression models for claim frequency prediction with telematics car driving information. In this paper, we propose two data-driven neural network approaches that process telematics car driving data to complement classical actuarial pricing with a driving behavior risk factor from telematics data. Our neural networks simultaneously accommodate feature engineering and regression modeling which allows us to integrate telematics car driving data in a one-step approach into the claim frequency regression models. We conclude from our numerical analysis that both classical actuarial risk factors and telematics car driving data are necessary to receive the best predictive models. This emphasizes that these two sources of information interact and complement each other.

Prediction of Solar Irradiance and Photovoltaic Solar Energy Product Based on Cloud Coverage Estimation Using Machine Learning Methods

Cloud cover estimation from images taken by sky-facing cameras can be an important input for analyzing current weather conditions and estimating photovoltaic power generation. The constant change in position, shape, and density of clouds, however, makes the development of a robust computational method for cloud cover estimation challenging. Accurately determining the edge of clouds and hence the separation between clouds and clear sky is difficult and often impossible. Toward determining cloud cover for estimating photovoltaic output, we propose using machine learning methods for cloud segmentation. We compare several methods including a classical regression model, deep learning methods, and boosting methods that combine results from the other machine learning models. To train each of the machine learning models with various sky conditions, we supplemented the existing Singapore whole sky imaging segmentation database with hazy and overcast images collected by a camera-equipped Waggle sensor node. We found that the U-Net architecture, one of the deep neural networks we utilized, segmented cloud pixels most accurately. However, the accuracy of segmenting cloud pixels did not guarantee high accuracy of estimating solar irradiance. We confirmed that the cloud cover ratio is directly related to solar irradiance. Additionally, we confirmed that solar irradiance and solar power output are closely related; hence, by predicting solar irradiance, we can estimate solar power output. This study demonstrates that sky-facing cameras with machine learning methods can be used to estimate solar power output. This ground-based approach provides an inexpensive way to understand solar irradiance and estimate production from photovoltaic solar facilities.

Hemoglobin and glucose level estimation from PPG characteristics features of fingertip video using MGGP-based model

Hemoglobin and the glucose level can be measured after taking a blood sample using a needle from the human body and analyzing the sample, the result can be observed. This type of invasive measurement is very painful and uncomfortable for the patient who is required to measure hemoglobin or glucose regularly. However, the non-invasive method only needed a bio-signal (image or spectra) to estimate blood components with the advantages of being painless, cheap, and user-friendliness. In this work, a non-invasive hemoglobin and glucose level estimation model have been developed based on multigene genetic programming (MGGP) using photoplethysmogram (PPG) characteristic features extracted from fingertip video captured by a smartphone. The videos are processed to generate the PPG signal. Analyzing the PPG signal, its first and second derivative, and applying Fourier analysis total of 46 features have been extracted. Additionally, age and gender are also included in the feature set. Then, a correlation-based feature selection method using a genetic algorithm is applied to select the best features. Finally, an MGGP based symbolic regression model has been developed to estimate hemoglobin and glucose level. To compare the performance of the MGGP model, several classical regression models are also developed using the same input condition as the MGGP model. A comparison between MGGP based model and classical regression models have been done by estimating different error measurement indexes. Among these regression models, the best results (±0.304 for hemoglobin and ±0.324 for glucose) are found using selected features and symbolic regression based on MGGP.

Transition models for count data: a flexible alternative to fixed distribution models

A flexible semiparametric class of models is introduced that offers an alternative to classical regression models for count data as the Poisson and Negative Binomial model, as well as to more general models accounting for excess zeros that are also based on fixed distributional assumptions. The model allows that the data itself determine the distribution of the response variable, but, in its basic form, uses a parametric term that specifies the effect of explanatory variables. In addition, an extended version is considered, in which the effects of covariates are specified nonparametrically. The proposed model and traditional models are compared in simulations and by utilizing several real data applications from the area of health and social science.

A fuzzy penalized regression model with variable selection

In the classical multiple regression modeling, there might be some insignificant input variables. These variables can be eliminated by automatic selectors, known as penalized methods. We propose a penalized estimation method for the coefficients of a linear regression model for studying the dependence of a LR fuzzy response (output) variable on a set of crisp explanatory (input) variables. To show the performances of the proposed model a simulation study was utilized under three scenarios of multicollinear and sparse data. The model demonstrates better performances in comparison to another three models on the basis of specified goodness of fit measures, under three variants of the penalty function. Evaluation of the method has been conducted on real data. The results demonstrate superior performances in terms of the goodness of fit measures in comparison to the other models. To take into account the imprecision due to the lack of knowledge about the data generation process, in the applications to real data bootstrap-t confidence intervals were also utilized.

Quality assessment of multiply and singly distorted stereoscopic images via adaptive construction of cyclopean views

A challenging problem confronted when designing a blind/no-reference (NR) stereoscopic image quality assessment (SIQA) algorithm is to simulate the quality assessment (QA) behavior of the human visual system (HVS) during binocular vision. An effective way to solve this problem is to estimate the quality of the merged single view created in the human brain which is also referred to as the cyclopean image. However, due to the difficulty in modeling the binocular fusion and rivalry properties of the HVS, obtaining effective cyclopean images for QA is non-trivial, and consequently previous NR SIQA algorithms either require the MOS/DMOS values of the distorted 3D images for training or ignore the quality analysis of the merged cyclopean view. In this paper, we focus on (1) constructing accurate and appropriate cyclopean views for QA of stereoscopic images by adaptively analyzing the distortion information of two monocular views, and (2) training NR SIQA models without requiring the assistance of the MOS/DMOS values in existing databases. Accordingly, we present an effective opinion-unaware SIQA algorithm called MUSIQUE-3D, which blindly assesses the quality of multiply and singly distorted stereoscopic images by analyzing quality degradations of both monocular and cyclopean views. The monocular view quality is estimated by an extended version of the MUSIQUE algorithm, and the cyclopean view quality is computed from the distortion parameter values predicted by a two-layer classification-regression model trained on a large 3D image dataset. Tests on various 3D image databases demonstrate the superiority of our method as compared with other state-of-the-art SIQA algorithms.

A Novel Machine Learning Algorithm Predicts Dementia With Lewy Bodies Versus Parkinson's Disease Dementia Based on Clinical and Neuropsychological Scores.

Our aim was to develop a machine learning algorithm based only on non-invasively clinic collectable predictors, for the accurate diagnosis of these disorders. This is an ongoing prospective cohort study (ClinicalTrials.gov identifier NCT number NCT04448340) of 78 PDD and 62 DLB subjects whose diagnostic follow-up is available for at least 3 years after the baseline assessment. We used predictors such as clinico-demographic characteristics, 6 neuropsychological tests (mini mental, PD Cognitive Rating Scale, Brief Visuospatial Memory test, Symbol digit written, Wechsler adult intelligence scale, trail making A and B). We investigated logistic regression, K-Nearest Neighbors (K-NNs) Support Vector Machine (SVM), Naïve Bayes classifier, and Ensemble Model for their ability to predict successfully PDD or DLB diagnosis. The K-NN classification model had an accuracy 91.2% of overall cases based on 15 best clinical and cognitive scores achieving 96.42% sensitivity and 81% specificity on discriminating between DLB and PDD. The binomial logistic regression classification model achieved an accuracy of 87.5% based on 15 best features, showing 93.93% sensitivity and 87% specificity. The SVM classification model had an accuracy 84.6% of overall cases based on 15 best features achieving 90.62% sensitivity and 78.58% specificity. A model created on Naïve Bayes classification had 82.05% accuracy, 93.10% sensitivity and 74.41% specificity. Finally, an Ensemble model, synthesized by the individual ones, achieved 89.74% accuracy, 93.75% sensitivity and 85.73% specificity. Machine learning method predicted with high accuracy, sensitivity and specificity PDD or DLB diagnosis based on non-invasively and easily in-the-clinic and neuropsychological tests.

Prediction of adverse drug reactions based on knowledge graph embedding

BackgroundAdverse drug reactions (ADRs) are an important concern in the medication process and can pose a substantial economic burden for patients and hospitals. Because of the limitations of clinical trials, it is difficult to identify all possible ADRs of a drug before it is marketed. We developed a new model based on data mining technology to predict potential ADRs based on available drug data.MethodBased on the Word2Vec model in Nature Language Processing, we propose a new knowledge graph embedding method that embeds drugs and ADRs into their respective vectors and builds a logistic regression classification model to predict whether a given drug will have ADRs.ResultFirst, a new knowledge graph embedding method was proposed, and comparison with similar studies showed that our model not only had high prediction accuracy but also was simpler in model structure. In our experiments, the AUC of the classification model reached a maximum of 0.87, and the mean AUC was 0.863.ConclusionIn this paper, we introduce a new method to embed knowledge graph to vectorize drugs and ADRs, then use a logistic regression classification model to predict whether there is a causal relationship between them. The experiment showed that the use of knowledge graph embedding can effectively encode drugs and ADRs. And the proposed ADRs prediction system is also very effective.

Logistic Regression: A Step by Step Solution Using Microsoft Excel

While knowledge of statistical and predictive analytical software is valued by the business community, it is assumed that business students have had extensive hands-on experience with Microsoft Excel and can be immediately productive with spreadsheets when walking in for their first day of work! This tutorial provides additional Microsoft Excels skills by showing how to create a logistic regression classification model, used in many business fields, from an intuitive perspective. The Solver nonlinear optimization Microsoft Excel add-in is used to derive the maximum likelihood estimates of the model coefficients. Model validation and performance are also completed with Microsoft Excel. Students are exposed to a wider view of optimization, and why it is at the heart of most machine learning algorithms. Student learning assessments from undergraduate and graduate classes are included to support our findings.

The comparison of the ordinal logistic model with the classical regression model

This research’s aims was to provide an alternative for statistical user who were interested in analysing the functional relationship between ordinal response and explanatory variable. The data used was the data of Students on Generation 2017 JPMIPA FKIP Bengkulu University. The test statistic from the modified probability ratio in ordinal logistic had a higher test power than the F test in the classical linear regression. Overall the probability of rejection of ordinal logistic tended to be lower than the classical regression. At condition of p close to one, the test power of the ordinal logistic would be the same as the classical regression. Ordinal logistic model tended to be more selective in choosing variables that significantly affect the response, resulting in a simpler model than the classical regression. The results of the research with the classical linear model obtained variables that influence the student GPA were UN score, High School Report, UNIB admission pathway, High School Status and interaction of UN Score and High School Status. While the ordinal logistic model produced influential independent variables were the UN Score, High School Report, High School Status and the interaction of UN Score and High School Status.

Exploring spatial dependence of cowpea-beans yield using global and local autocorrelation statistics in the Eastern Cariri region of Paraíba

ABSTRACT: This study evaluated the variability and characterizedthe spatial dependence between some soil attributes in the Eastern Cariri microregion of Paraíba,and analyzed the spatial correlations in order to identify the interactions between such attributes in cowpea bean(Vigna unguiculata L. Walp)production. Harvest data of the agricultural years of 2000-2017 in the Eastern Cariri microregion of Paraíba were analyzed. Parameters of the fitted models wereestimated using the Maximum Likelihood method and the performance of the models was evaluated based on coefficients of determination(R2), maximum log-likelihood function, and Schwarz’s Bayesian information criterion (BIC). Correlation and spatial autocorrelation between the cowpea productivity and agrometeorological elements was detected through the spatial analysis, using techniques such as the Moran’s index I. The study showed that, according to the performance indicators used, the spatial error model offered better results in relation to the classical multiple regression models and the self-regressive spatial models, indicating that the inclusion of spatial dependence in the models improves the estimate of productivity of cowpea in the microregion of Cariri Oriental da Paraíba.

Guiding iterative optimisation methods to a predefined kind of optima for unconstrained optimisation problems

One of the most fundamental issues in the field of mathematical optimisation is the convergence of an iterative optimisation method and by this we are referring to two things. First, will the method find an optimum and second, will this optimum be a local one or a global one? A recently proposed technique (Nikolakakou et al., 2015b) that is used in order to lessen the dependence a locally convergent iterative optimisation method has on the initial guess, is exploited in this paper. A way so that such a method may be guided to a predefined kind of minimum (local or global) is presented. Well known test functions were used for experimentation. Statistical analysis was conducted by applying a logistic regression classification model on data arisen from the numerical results which confirmed that iterative optimisation methods can be guided to a predefined kind of optimum.

Social Media and Stock Market Prediction: A Big Data Approach

Big data is the collection of large datasets from traditional and digital sources to identify trends and patterns. The quantity and variety of computer data are growing exponentially for many reasons. For example, retailers are building vast databases of customer sales activity. Organizations are working on logistics financial services, and public social media are sharing a vast quantity of sentiments related to sales price and products. Challenges of big data include volume and variety in both structured and unstructured data. In this paper, we implemented several machine learning models through Spark MLlib using PySpark, which is scalable, fast, easily integrated with other tools, and has better performance than the traditional models. We studied the stocks of 10 top companies, whose data include historical stock prices, with MLlib models such as linear regression, generalized linear regression, random forest, and decision tree. We implemented naive Bayes and logistic regression classification models. Experimental results suggest that linear regression, random forest, and generalized linear regression provide an accuracy of 80%–98%. The experimental results of the decision tree did not well predict share price movements in the stock market.

Deep Learning under Model Uncertainty

Deep learning has proven to lead to very powerful predictive models, often outperforming classical regression models such as generalized linear models. Deep learning models perform representation learning, which means that they do covariate engineering themselves so that explanatory variables are optimally transformed for the predictive problem at hand. A crucial object in deep learning is the loss function (objective function) for model fitting which implicitly reflects the distributional properties of the observed samples. The purpose of this article is to discuss the choice of this loss function, in particular, we give a specific proposal of a loss function choice under model uncertainty. This proposal turns out to robustify representation learning and prediction.