Search

This study aims to profile the scientific retractions published in journals indexed in the Web of Science database from 2010 to 2019, from researchers at the top 20 World Class Universities according to the Times Higher Education global ranking of 2020. Descriptive statistics, Pearson's correlation coefficient, and simple linear regression were used to analyze the data. Of the 330 analyzed retractions, Harvard University had the highest number of retractions and the main reason for retraction was data results. We conclude that the universities with a higher ranking tend to have a lower rate of retraction.

The formalism is proposed for assessing the accuracy of simple regression takes into account both the correlation of the Y variable data and the impact of type B uncertainty in routine measurements.This is the continuation of authors’ work considering the data with type B uncertainty and uncorrelated random errors. The essence, criteria and dependencies of the regression method were examined. Simulated examples of determining uncertainty bands of the regression line fitted to measured points with different cases of correlated values of dependent Y variable are considered. The recommendations of GUM Guide [1] was referred to and the type B not discussed yet in the literature was considered. The proposed formalism is illustrated by examples the precisely know abscissa and ordinates with different correlation, and absolute and relative uncertainties type A and type B.

Kernel frequency distributions consist of multiple identical Gaussian curves rather than histograms consistent of bins with different lengths. Kernel regression measures the relationship between x and y data, where the expectation of y is conditional not on all x-values but on locally weighted averages of subsets of consecutive x-values within a bandwidth. And, together, they tend to produce a simple linear regression model. Any linear regression can be replaced with a kernel regression, but the method is particularly appropriate in case of seemingly nonlinear patterns, that, in the end, are linear after all. With skewed data files a (much) better fit for the data is provided by kernel regression than by linear regression.

In this chapter, a complete overview of prediction and classification processes has been provided. The concept of prediction is explained with simple linear regression examples that are easily and manually solved; with this, in-depth idea of regression is understandable for readers. In continuation to this, multiple linear regression is also explained. After this, the concept of classification has been explained through logistic regression. Complete process of feature extraction and classification has been demonstrated by real ECG dataset. MatLab codes have also been given to get the real experience of predication and classification processes. After getting an idea of regression and logistic function used for classification, the concept of artificial neural networks becomes very easy to understand. Basic and necessary steps of ANN and its training and testing have been explained. After reading this chapter and following the given examples, readers will easily understand the complex models of ANN and deep neural networks for the classification and prediction problems.

Algorithm selection refers to the task of automatically selecting the most suitable algorithm for solving an instance of a computational problem from a set of candidate algorithms. Here, suitability is typically measured in terms of the algorithms’ runtimes. To allow the selection of algorithms on new problem instances, machine learning models are trained on previously observed performance data and then used to predict the algorithms’ performances. Due to the computational effort, the execution of such algorithms is often prematurely terminated, which leads to right-censored observations representing a lower bound on the actual runtime. While simply neglecting these censored samples leads to overly optimistic models, imputing them with precise though hypothetical values, such as the commonly used penalized average runtime, is a rather arbitrary and biased approach. In this paper, we propose a simple regression method based on so-called superset learning, in which right-censored runtime data are explicitly incorporated in terms of interval-valued observations, offering an intuitive and efficient approach to handling censored data. Benchmarking on publicly available algorithm performance data, we demonstrate that it outperforms the aforementioned naive ways of dealing with censored samples and is competitive to established methods for censored regression in the field of algorithm selection.

Data obtained from low-cost environmental sensors can have various issues such as low precision and accuracy and incompleteness. A calibration process is often applied to address such issues. With the recent advances in artificial intelligence, we have seen an increased number of applications that starts to use an artificial neural network (ANN) to calibrate the sensors, and their results are promising. In this work, we used a six-months worth of real hourly data to demonstrate that the ANN may not always be the best choice of a calibration method. Our evaluation compares an ANN-based method with a simple regression-based method in various aspects. The result shows that the ANN-based method does not consistently outperform the regression-based method. More interestingly, in the comparison, our results suggest that the performance of a calibration can be more sensitive to some of the factors (e.g. training and testing data, model parameters) than the use of different calibration methods. Even though the results may not be generalised in other sensors or datasets, our evaluation provides evidence showing that inappropriate use of a calibration method can compromise the calibration result, and the use of the ANN will not magically solve that problem.

This study examines the determinants of planned retirement age of informal workers in Chiang Mai Province, Thailand. The least absolute shrinkage and selection operator estimator regression (LASSO) and the ordinary least squares estimator regression (OLS) are employed to determine variables important for planned retirement age equation. The computational cost of this method is we need to take into account 2 times fewer variables compared with standard approaches in the simple regression literature. A cross-sectional study is conducted among informal workers in Chiang Mai Province, Thailand. A total of 400 informal workers are enrolled. A number of important participants, demographic characteristics, financial variables and health factors are considered in the planned retirement age equation. The study compares the results of two methods, namely the LASSO and the OLS. According to the minimum value of The out-of-sample Root Mean Squared Error, the LASSO performs better than the OLS. The results suggest that age of respondent, number of children, own accommodation, sex of respondent, and marital status (single) have positive impact on planned retirement age. In contrast, cost of bone disease treatment, daily expenses, caretaker after retirement, chronic disease, education level, cost of treatment for diseases, income after retirement, financial burden after retirement, and savings group deposits have negative impact on planned retirement age.

In the big data era, data scientists explore machine learning methods for observed data to predict or classify. For machine learining to be effective, it requires access to raw data which is often privacy sensitive. In addition, whatever data and fitting procedures are employed, a crucial step is to select the most appropriate model from the given dataset. Model selection is a key ingredient in data analysis for reliable and reproducible statistical inference or prediction. To address this issue, we develop new techniques to provide solutions for running model selection over encrypted data. Our approach provides the best approximation of the relationship between the dependent and independent variable through cross validation. After performing 4-fold cross validation, 4 different estimates of our model’s errors are calculated. And then we use bias and variance extracted from these errors to find the best model. We perform an experiment on a dataset extracted from Kaggle and show that our approach can homomorphically regress a given encrypted data without decrypting it.

This chapter gives suggestions and guidelines for choosing statistical tests on the basis of study design (univariate, bivariate, multivariate), level of measurement, and distribution of the data in the population. First it deals with analyses of relationships between two variables (bivariate) by means of parametric tests (e.g., t-test, ANOVA, Pearson’s correlation coefficient, simple linear regression) when the data are interval or continuous, normally distributed, and randomly drawn from the population or using non-parametric tests (e.g., χ2, Mann-Whitney, Wilcoxon signed-rank, Kruskal-Wallis, and Spearman’s rank correlation) for nominal or ordinal data and small (

This chapter explains the basics of simple linear regression. Showing different approaches to solve a prediction problem of this type. First, we explain the theory, then, it is solved by the algebraic method of least squares, checking the procedure and results through MATLAB. Finally, a basic example of the field of evolutionary computing is shown, using several evolutionary techniques such as PSO, DE, ABC, CS and the classical method of descending gradient. Which optimize the function to find the best coefficients for an estimated straight line. This is applied to a set of fatal traffic accident data in the U.S. states.

Decision making is a process of reaction against organizational hazards and opportunities. It includes the process of collecting and processing the information gathered and selecting the alternative from the set of alternatives based on their values using different tools, techniques, and insights. Regression analysis is one among the most dominant techniques used for decision making in business by the management. To make better decisions, regression analysis helps the managers to understand the data to model dependencies and helps to understand the relationship between the expected output and the input features to predict the values. The main application of regression analysis is to find how strong an independent variable influence the dependent variable. There are many areas where regression analysis can be applied in organizations for better prediction, mainly in financial forecasting, marketing, understanding inventory levels, supply chain, trend analysis, and time series prediction. In this paper, the application of regression analysis in different organizational decision making and the different types of regressions used in organizational decision making are discussed.

While deep learning-based classification is generally tackled using standardized approaches, a wide variety of techniques are employed for regression. In computer vision, one particularly popular such technique is that of confidence-based regression, which entails predicting a confidence value for each input-target pair (x, y). While this approach has demonstrated impressive results, it requires important task-dependent design choices, and the predicted confidences lack a natural probabilistic meaning. We address these issues by proposing a general and conceptually simple regression method with a clear probabilistic interpretation. In our proposed approach, we create an energy-based model of the conditional target density p(y|x), using a deep neural network to predict the un-normalized density from (x, y). This model of p(y|x) is trained by directly minimizing the associated negative log-likelihood, approximated using Monte Carlo sampling. We perform comprehensive experiments on four computer vision regression tasks. Our approach outperforms direct regression, as well as other probabilistic and confidence-based methods. Notably, our model achieves a \(2.2\%\) AP improvement over Faster-RCNN for object detection on the COCO dataset, and sets a new state-of-the-art on visual tracking when applied for bounding box estimation. In contrast to confidence-based methods, our approach is also shown to be directly applicable to more general tasks such as age and head-pose estimation. Code is available at https://github.com/fregu856/ebms_regression.

Approximating the solution value of transportation problems has become more relevant in recent years, as these approximations can help to decrease the computational effort required for solving those routing problems. In this paper, we apply several regression methods to predict the total distance of the traveling salesman problem (TSP) and vehicle routing problem (VRP). We show that distance can be estimated fairly accurate using simple regression models and only a limited number of features. We use features found in scientific literature and introduce a new class of spatial features. The model is validated on a dynamic waste collection case in the city of Amsterdam, the Netherlands. We introduce a cost function that combines the travel distance and service level, and show that our model can reduce distances up to 17%, while maintaining the same service level, compared to a well-known heuristic approximation. Furthermore, we show the benefits of using approximations for combining offline learning with online or frequent optimization.

Big Data analytics is the latest emerging technology that requires deep knowledge in business intelligence, machine learning, and statistical methods and in deep learning. It focuses on the application of data analytics for energy demand management using real-time data. The data is then analyzed for clustering, demand forecasting, pricing, and energy generation optimization. It represents a method to predict energy usage, based on real-time data obtained from TANGEDCO-CBE, using the linear regression model (LR). The final linear regression models developed were based on daily sustained demand and consumption by comparing actual and predicted energy usage models can predict with acceptable errors. Normally the energy requirement and industrial demands are high; hence the application of these energy Big Data analyses significantly improves efficiency and provides new business opportunities.

Gaussian Process (GP) regression has seen widespread use in robotics due to its generality, simplicity of use, and the utility of Bayesian predictions. The predominant implementation of GP regression is a nonparametric kernel-based approach, as it enables fitting of arbitrary nonlinear functions. However, this approach suffers from two main drawbacks: (1) it is computationally inefficient, as computation scales poorly with the number of samples; and (2) it can be data inefficient, as encoding prior knowledge that can aid the model through the choice of kernel and associated hyperparameters is often challenging and unintuitive. In this work, we propose ALPaCA, an algorithm for efficient Bayesian regression which addresses these issues. ALPaCA uses a dataset of sample functions to learn a domain-specific, finite-dimensional feature encoding, as well as a prior over the associated weights, such that Bayesian linear regression in this feature space yields accurate online predictions of the posterior predictive density. These features are neural networks, which are trained via a meta-learning (or “learning-to-learn”) approach. ALPaCA extracts all prior information directly from the dataset, rather than restricting prior information to the choice of kernel hyperparameters. Furthermore, by operating in the weight space, it substantially reduces sample complexity. We investigate the performance of ALPaCA on two simple regression problems, two simulated robotic systems, and on a lane-change driving task performed by humans. We find our approach outperforms kernel-based GP regression, as well as state of the art meta-learning approaches, thereby providing a promising plug-in tool for many regression tasks in robotics where scalability and data-efficiency are important.

This paper shows how students can be guided to integrate elementary mathematical analyses with motion planning for typical educational robots. Rather than using calculus as in comprehensive works on motion planning, we show students can achieve interesting results using just simple linear regression tools and trigonometric analyses. Experiments with one robotics platform show that use of these tools can lead to passable navigation through dead reckoning even if students have limited experience with use of sensors, programming, and mathematics.

Contract farming has a managerial advantage that farmers can directly negotiate prices with business partners. And it is necessary to predict harvesting date of agricultural crops precisely for contract farming. Conventionally, this precondition has been solved by experience rule of experienced farmers or some mathematical model and simple regression models. However, for new farmers who do not have experiential rules, contract farming is a difficult management method. And previous prediction models cannot consider complex relationship between environmental parameters. In order to solve this problem, this research proposed automatic harvesting date prediction method using statistical model by deep learning. This research confirmed that deep learning models exceed the accuracy of non-deep learning models. And this research proposed a practical system using our models and one existing Web application for management of farming tasks.

It is well known that there is a strong relationship between the amount of precipitation and radial growth of the trees, but it is not known up to which extent they depends on each other. On the other hand, the data about the amount of precipitation is available from 1976 in Duhok governorate in general. One of the problems studied here is to estimate the quantity of the precipitation using the width of annual ring for the periods prior to 1976, in order to see the trend of rain and snow fall in the region. The data used in this study came from 387 sample pairs of precipitation and diameter growth. These data were undergone data processing for the purpose of developing of regression equations between the width of diameter growth as dependent variable and the amount of yearly precipitation as independent variables in regression equations. Accordingly, thirteen regression equations were developed; two of them were simple regression equations, one polynomial, and the rest of equations were nonlinear. These equations were undergone several measures of precision for the purpose of selecting the most appropriate one which fits our data set. Ultimately the non-linear regression equation: Dg = 0.293781 + 0.0000371 × p1.40698 was selected, with an adjusted R2 of 76.22, standard error of estimate of 0.1283 and DW of 1.88 The selected equation can be used to estimate the amount of radial growth of a tree in the region by substituting the amount of precipitation in the equation.

The rapid development and deployment of Intelligent Transportation Systems (ITSs) require the development of data driven algorithms. Travel time modeling is an integral component of travel and transportation management and travel demand management functions. Travel time has a massive impact on driver’s route choice behavior and the assessment of the transportation system performance. In this paper, a mixture of linear regression is proposed to model travel times. The mixture of linear regression models has three advantages. First, it provides better model fitting compared to simple linear regression. Second, the proposed model can capture the bi-modal nature of travel time distributions and link it to the uncongested and congested traffic regimes. Third, the means of the bi-modal distributions are modeled as functions of the input predictors. This last advantage allows for the quantitative evaluation of the probability of each travel time state as well as the uncertainty associated with each state at any time of the day given the values of the predictors at that time. The proposed model is applied to archived data along a 74.4-mile freeway stretch of I-66 eastbound to connect I-81 and Washington D.C. The experimental results show the ability of the model to capture the stochastic nature of the travel time and gives good travel time predictions.

The use of predictive models at the border inspection posts in a port may help to manage and plan operations processes in such a way that time delays and congestion issues are minimized. In this paper, an enriched time series database containing records of the number of inspections carried out in the Port of Algeciras Bay between 2010 and 2018 is analyzed using two well-known statistical and computational intelligence methods such as linear regression (baseline model) and deep-fully connected neural networks. Additionally, a deep ensemble neural network approach is proposed in order to try to boost predictive performance even further. The results of the analysis show how deep fully-connected neural networks outperform a simple linear regression model, in particular the ensemble approach obtains performances of \(\sigma =0.813\) and \(MSE=330.160\) in contrast to \(\sigma =0.804\) and \(MSE=342.721\) achieved by linear regression. A visual comparison of the original and predicted time series shows how the ensemble approach is able to model better high and low peaks than the time series predicted by linear regression.

Disease prediction based on modeling the correlations between compounded indicator factors is a widely used technique in high incidence chronic disease prevention diagnosis. Predictive models based on personal health information have been developed historically by using simple regression fitting over relatively few factors. Regression approaches have been favored in previous prediction modeling approaches because they are simplest and do not assume any non-linearity in the model for contributions of the chosen factors. In practice, many factors are correlated and have underlying non-linear relationships to the predicted outcome. Deep learning offers a means to construct a more complex modeling approach, along with automation and adaptation. The aim of this paper is to assess the ability of a deep learning model to predict the heart disease incidence using a common benchmark dataset (University of California, Irvine (UCI) dataset). The performance of deep learning model has been compared with four popular machine learning models (two linear and two nonlinear) in predicting the incidence of heart disease using data from 567 participants from two cohorts taken from UCI database. The deep learning model was able to achieve the best accuracy of 94% and an AUC score of 0.964 when compared to other models. The performance of deep learning and nonlinear machine learning models was significantly better compared to the linear machine learning models with increase in the dataset size.

Deep learning generates state-of-the-art semantic segmentation provided that a large number of images together with pixel-wise annotations are available. To alleviate the expensive data collection process, we propose a semi-supervised domain adaptation method for the specific case of images with similar semantic content but different pixel distributions. A network trained with supervision on a past dataset is finetuned on the new dataset to conserve its features maps. The domain adaptation becomes a simple regression between feature maps and does not require annotations on the new dataset. This method reaches performances similar to classic transfer learning on the PASCAL VOC dataset with synthetic transformations.

This research aims to examine factors affecting Jordanian consumers’ attitude towards Facebook advertising (entertainment, informativeness, irritation, credibility, peer influence and privacy concerns). In order to test the proposed conceptual framework, an online web-based survey was employed, and data was collected from 380 university students in Jordan. Simple linear regression and multiple regression analysis, using the Statistical Package for the Social Sciences (SPSS) version 17, were employed to analyse the collected data. Results showed that entertainment, informativeness, interactivity, credibility and privacy concerns have a direct positive effect on Jordanian consumers’ attitude towards Facebook advertising in tourism. However, entertainment has the most significant effect, while credibility has the lowest significant effect. The study recommends tourism companies to take into consideration the importance of these variables when designing their ads on Facebook in order to be able to benefit from this huge new virtual world of marketing opportunities in an ethical way.

In recent years, call detail records (CDRs) have been widely used to study various aspects of urban and human dynamics. One assumption implicitly made in many existing studies is that people’s phone communication activities could represent spatiotemporal distribution of the population, or at least of the mobile phone users. By using a mobile phone data set which consists of CDRs plus other cellphone-related logs (e.g., cellular handover and periodic location update), we derive two cellphone usage indicators (volume of calls/messages [\(V\)] and number of active phone users [\(N\)]) as well as the spatiotemporal distribution of mobile phone users, and evaluate their relationships through correlation and regression analysis. We find that the correlations between the number of mobile phone users and each of the two cellphone usage indicators remain high and stable during the day time and in early evening (i.e., 07:00–21:30). However, their relationships revealed by the regression models vary greatly throughout a day. Researchers therefore should be cautious when using mobile phone communication activities to quantify certain aspects of urban dynamics. Our regression analyses suggest that the log-transformation model performs better than the simple linear regression model (in predicting phone user distribution) when the independent variable (\(V\) or \(N\)) is fixed. Also, we find that \(N\) serves as a better independent variable than \(V\), which is affected more by individual “burst” of phone communication activities, when explaining spatiotemporal distribution of mobile phone users. A 3-fold cross validation suggests that CDRs can be used along with other data sources (e.g., land use) to deliver more robust estimation of phone user distributions, which potentially facilitate dynamic projection of urban population distributions.

This paper aims to predict energy demand in Spain for the year 2020 and analyzes whether this country will be able to meet the European Union’s greenhouse gas emission reduction commitment. To this purpose, we use climatic data and some variables to measure the economic activity in Spain. The simulated scenario considers that Spain will begin a process of economic recovery which will result in an increase in industrial activity with stable climatic conditions. Several techniques including Simple Linear Regression, Support Vector Machines or Deep Learning have been proposed to estimate and test the model. The EU agreements imply that by 2020 between 20 and 30% of the consumed energy will come from clean and renewable energy sources. The conclusions for this paper show that Spain may be on track to meet its commitments to Europe.

The aesthetic quality assessment of images is a challenging work in the field of computer vision because of its complex subjective semantic information. The recent research work can utilize the deep convolutional neural network to evaluate the overall score of the image. However, the focus in the field of aesthetic is often not limited to the total score of image, and multiple attribute of the aesthetic evaluation can obtain image richer aesthetic characteristics. The multi-attribute rating called Aesthetic Radar Map. In addition, traditional deep learning methods can only be predicted by classification or simple regression, and cannot output multi-dimensional information. In this paper, we propose a hierarchical multi-task dense network to make multiple regression of the properties of images. According to the total score, the scoring performance of each attribute is enhanced, and the output effect is better by optimizing the network structure. Through this method, the more sufficient aesthetic information of the image can be obtained, which is of certain guiding significance to the comprehensive evaluation of image aesthetics.

EN 9100 is a quality management system standard for the aerospace industry derived from the ISO 9000 standard. The aerospace industry is economically prominent, both worldwide and in Spain. The goals of this paper are (a) to analyze the motivations of Spanish aerospace firms in adhering to EN 9100 standard and (b) to examine whether the type of motivation affects the results of implementing this standard. To accomplish this, both ANOVA and a simple linear regression model were applied to data from the 122 aerospace industry valid survey responses. The results demonstrate that most firms adhere to EN 9100 in the Spanish aerospace industry due to “external” motivations, such as to increase their institutional legitimacy and reputation. Nevertheless, firms, where “internal” motivations such as to improve their operational execution or organizational processes are predominant, showed superior benefits as a result of implementing the standards. The conclusions of this article may be of interest both for academic and professional spheres of activity.

In this paper, we introduce the holdout sampler to find the defective pathways in high underdetermined phenotype prediction problems. This sampling algorithm is inspired by the bootstrapping procedure used in regression analysis to established confidence bounds. We show that working with partial information (data bags) serves to sample the linear uncertainty region in a simple regression problem, mainly along the axis of greatest uncertainty that corresponds to the smallest singular value of the system matrix. This procedure applied to a phenotype prediction problem, considered as a generalized prediction problem between the set of genetic signatures and the set of classes in which the phenotype is divided, serves to unravel the ensemble of altered pathways in the transcriptome that are involved in the disease development. The algorithm looks for the minimum-scale genetic signature in each random holdout and the likelihood (predictive accuracy) is established using the validation dataset via a nearest-neighbor classifier. The posterior analysis serves to identify the header genes that most-frequently appear in the different hold-outs and are therefore robust to a partial lack of samples. These genes are used to establish the genetic pathways and the biological processes involved in the disease progression. This algorithm is much faster, robust and simpler than Bayesian Networks. We show its application to a microarray dataset concerning a type of breast cancers with poor prognoses (TNBC).

In previous chapters, we have had data for which there has been a dependent variable (Y ) and an independent variable (X – even though, to be consistent with the notation that is close to universal in the field of experimental design, we have been using factor names, A, B, etc., or “column factor” and “row factor,” instead of, literally, the letter X ). The latter has been treated mostly as a categorical variable, whether actually numerical/metric or not. Often, we have had more than one independent variable. Assuming only one independent variable, if we want to say it this way (and we do!), we can say that we have had n (X, Y ) pairs of data, where n is the total number of data points. With more than one independent variable, we can say that we have n (X1, X2, …, Y ) data points.