Search

Abstract The researchers in the current era provided many new recommendation methodologies. Though various recommendation techniques exist, there is a need to develop a unique technique for capturing latent factors and patterns from sparse and high-dimensional data in pervasive environments, specifically for optimizing dynamic recommendations. This study proposes a hybrid approach for optimizing dynamic recommendations in pervasive environments by combining Non-Negative Matrix Factorization (NMF) with deep learning and reinforcement learning techniques. The goal is to overcome the challenge of capturing latent factors and patterns from sparse and high-dimensional data. By leveraging NMF, meaningful latent factors are extracted, while deep learning, specifically Faster recurrent neural networks (FRNNs), learns complex feature representations. Reinforcement learning algorithms optimize the recommendation policy based on user feedback. This Hybrid Context-Aware Optimized Recommendation (HCOR) approach improves recommendation accuracy and relevance in pervasive environments, adapts to changing contexts, and enhances user experiences. The performance benefits are achieved by effectively capturing latent factors and patterns, resulting in improved accuracy and the ability to provide personalized and context-aware recommendations. The performance indicators to validate the research work include the recommendations' accuracy, relevance, and adaptability in pervasive environments. Additionally, metrics, such as precision, recall, and F1-score, are used to evaluate the effectiveness of the hybrid approach in capturing latent factors and patterns. User feedback and satisfaction are also measured to assess the impact on user experiences. The HCOR approach shows substantial performance gains, measuring a precision of 0.932, a recall of 0.922, and an F1-score of 0.943, which indicates the excellent ability of the approach to deliver accurate and personalized recommendations in a pervasive environment.

Estimating the number of latent factors underlying a set of dichotomous items is a major challenge in social and behavioral research. Mokken scale analysis (MSA) and exploratory graph analysis (EGA) are approaches for partitioning measures consisting of dichotomous items. In this study we perform simulation-based comparisons of two EGA methods (EGA with graphical least absolute shrinkage and selector operator; EGAtmfg with triangulated maximally filtered graph algorithm), two MSA methods (AISP: automated item selection procedure; GA: genetic algorithm), and two widely used factor analytic techniques (parallel analysis with principal component analysis (PApc) and parallel analysis with principal axis factoring (PApaf)) for partitioning dichotomous items. Performance of the six methods differed significantly according to the data structure. AISP and PApc had highest accuracy and lowest bias for unidimensional structures. Moreover, AISP demonstrated the lowest rate of misclassification of items. Regarding multidimensional structures, EGA with GLASSO estimation and PApaf yielded highest accuracy and lowest bias, followed by EGAtmfg. In addition, both EGA techniques exhibited the lowest rate of misclassification of items to factors. In summary, EGA and EGAtmfg showed comparable performance to the highly accurate traditional method, parallel analysis. These findings offer guidance on selecting methods for dimensionality analysis with dichotomous indicators to optimize accuracy in factor identification.

Estimating the number of latent factors underlying a set of dichotomous items is a major challenge in social and behavioral research. Mokken scale analysis (MSA) and exploratory graph analysis (EGA) are approaches for partitioning measures consisting of dichotomous items. In this study we perform simulation-based comparisons of two EGA methods (EGA with graphical least absolute shrinkage and selector operator; EGAtmfg with triangulated maximally filtered graph algorithm), two MSA methods (AISP: automated item selection procedure; GA: genetic algorithm), and two widely used factor analytic techniques (parallel analysis with principal component analysis (PApc) and parallel analysis with principal axis factoring (PApaf)) for partitioning dichotomous items. Performance of the six methods differed significantly according to the data structure. AISP and PApc had highest accuracy and lowest bias for unidimensional structures. Moreover, AISP demonstrated the lowest rate of misclassification of items. Regarding multidimensional structures, EGA with GLASSO estimation and PApaf yielded highest accuracy and lowest bias, followed by EGAtmfg. In addition, both EGA techniques exhibited the lowest rate of misclassification of items to factors. In summary, EGA and EGAtmfg showed comparable performance to the highly accurate traditional method, parallel analysis. These findings offer guidance on selecting methods for dimensionality analysis with dichotomous indicators to optimize accuracy in factor identification. [ABSTRACT FROM AUTHOR]

To elucidate the brain‐wide information interactions that vary and contribute to individual differences in schizophrenia (SCZ), an information‐resolved method is employed to construct individual synergistic and redundant interaction matrices based on regional pairwise BOLD time‐series from 538 SCZ and 540 normal controls (NC). This analysis reveals a stable pattern of regionally‐specific synergy dysfunction in SCZ. Furthermore, a hierarchical Bayesian model is applied to deconstruct the patterns of whole‐brain synergy dysfunction into three latent factors that explain symptom heterogeneity in SCZ. Factor 1 exhibits a significant positive correlation with Positive and Negative Syndrome Scale (PANSS) positive scores, while factor 3 demonstrates significant negative correlations with PANSS negative and general scores. By integrating the neuroimaging data with normative gene expression information, this study identifies that each of these three factors corresponded to a subset of the SCZ risk gene set. Finally, by combining data from NeuroSynth and open molecular imaging sources, along with a spatially heterogeneous mean‐field model, this study delineates three SCZ synergy factors corresponding to distinct symptom profiles and implicating unique cognitive, neurodynamic, and neurobiological mechanisms. [ABSTRACT FROM AUTHOR]

Overpasses are constructed because they allow continuous passage of pedestrians without disturbing the flow of vehicles. However, research from developing countries along with the anecdotal evidence from the study location revealed that generally most pedestrians prefer not using overpasses in crossing roads, rendering them inefficient and causing safety concerns. As such, this paper examines the factors - both observable and latent - influencing pedestrians' overpass utilization preference. The study was situated in Iligan City, Philippines, wherein four overpasses in the city were investigated by conducting on-site observations and questionnaire surveys. The data collected were analyzed using a combination of multiple linear regression (MLR), exploratory factor analysis (EFA), and structural equation modeling (SEM). On-site pedestrian traffic count revealed that the overpasses in Iligan City are generally ineffective, with only 38.42% average utilization rate. The MLR revealed three observable contributing factors that may affect pedestrian overpass crossing choice: having a driver's license, the overpass width, and the overpass span. EFA and SEM were able to identify safety, convenience, facility condition, and security as the latent factors having a positive direct influence on the preference of pedestrians overpass utilization. These results are instrumental at determining areas of concern relating to overpass design and improvements to increase the utilization rates of the overpass facilities in the city. [ABSTRACT FROM AUTHOR]

Purpose: Cost and time are two of the primary benchmarks in which construction projects are measured. A variety of factors, however, affect cost and time on construction projects, as identified in previous research. This has led to a need for better understanding how factors affecting cost and time overruns on public construction projects can be managed more efficiently. The purpose of this paper is to address these issues. Design/methodology/approach: In this study 26 factors affecting cost and time overruns on construction projects were identified, through qualitative interviews with project managers from Danish governmental agencies and through a literature review. Through principal component analyses the 26 factors were subsequently narrowed down to four primary latent factors. Findings: The identified four latent factors affecting cost and time overruns on public construction projects were lack of quality management, lack of project pre-planning, lack of user management and lack of project management. Originality/value: Previous research has focussed on increasing knowledge by identifying and ranking factors affecting time and cost performance. This has led to the identification of an overwhelming number of factors to use for managing construction projects. The present research reduced the number of factors by clustering them into key latent factors responsible for most of the deviation in performance, narrowing the scope of construction cost and time management into a few tangible key focus areas. This supports and improves fast decision-making that is necessary in a changeable environment such as construction. [ABSTRACT FROM AUTHOR]

E-scooters have emerged as a popular micromobility option for short trips, with many cities embracing shared e-scooters to enhance convenience for travelers and reduce reliance on automobiles. Despite their rising popularity, there is a lack of clear understanding of how user preferences and adoption practices vary by location. This study aims to explore user and non-user attitudes towards e-scooter use in diverse urban settings. A meta-analysis of data from three surveys (N = 1197) conducted in Washington, D.C., Miami, FL, and Los Angeles, CA, was performed to compare e-scooter users and non-user profiles, mode choice factors, and attitudes and preferences towards e-scooter use. Additionally, machine learning (ML) and SHAP (SHapley Additive exPlanations) analysis were utilized to identify influential factors in predicting e-scooter use in each city. The results reveal that the majority of e-scooter users are 25 to 39 of age, male, with higher income and a bachelor's degree, and 92% possess a driver's license. Significant differences in attitudes between e-scooter users and non-users highlight the complexity of perceptions towards e-scooter usage. The ML model indicates that employment status negatively impacts the prediction of e-scooter users, while factors such as living without a car and using non-motorized modes positively influence e-scooter use. Educational background is a significant e-scooter mode choice factor in Washington, D.C. and Miami, whereas attitudinal questions on car and technology usage are influential in Los Angeles. These findings provide valuable insights into the factors shaping e-scooter adoption, informing urban transportation planning and policymaking and enhancing understanding of shared micromobility and its impact on urban mobility. [ABSTRACT FROM AUTHOR]

Transgender and nonbinary people (TNB) in the U.S. experience high HIV prevalence and diverse economic hardships. Yet a comprehensive understanding of how multiple, simultaneously occurring hardships—termed economic marginality—are together associated with healthcare and HIV outcomes is needed. Leveraging survey data from a sample of 330 TNB people in three U.S. cities, we conducted an exploratory mixed-source principal component analysis of latent factors of economic experience, then estimated their associations with sexual behavior, access to healthcare, HIV status, and HIV testing frequency. Two factors emerged: a traditional socioeconomic factor related to income, education, and employment (SES), and one related to housing precarity and (lack of) assets (Precarity). Higher Precarity scores were associated with sexual behavior, cost-based healthcare avoidance, discrimination-based healthcare avoidance, and more frequent HIV testing. Findings highlight the importance of understanding profiles of economic marginalization among trans and nonbinary people and can inform efforts to address upstream, structural factors shaping healthcare access and HIV outcomes in this key population. [ABSTRACT FROM AUTHOR]

Abstract To elucidate the brain‐wide information interactions that vary and contribute to individual differences in schizophrenia (SCZ), an information‐resolved method is employed to construct individual synergistic and redundant interaction matrices based on regional pairwise BOLD time‐series from 538 SCZ and 540 normal controls (NC). This analysis reveals a stable pattern of regionally‐specific synergy dysfunction in SCZ. Furthermore, a hierarchical Bayesian model is applied to deconstruct the patterns of whole‐brain synergy dysfunction into three latent factors that explain symptom heterogeneity in SCZ. Factor 1 exhibits a significant positive correlation with Positive and Negative Syndrome Scale (PANSS) positive scores, while factor 3 demonstrates significant negative correlations with PANSS negative and general scores. By integrating the neuroimaging data with normative gene expression information, this study identifies that each of these three factors corresponded to a subset of the SCZ risk gene set. Finally, by combining data from NeuroSynth and open molecular imaging sources, along with a spatially heterogeneous mean‐field model, this study delineates three SCZ synergy factors corresponding to distinct symptom profiles and implicating unique cognitive, neurodynamic, and neurobiological mechanisms.

Objectives: Perinatal psychopathology can be damaging. This study examined the strength of the associations between risk factors and all perinatal mood and anxiety disorder symptoms while assessing the mediating effect of experiential avoidance. Method: Participants (N = 246) completed assessments during pregnancy (28–32 weeks) and the postpartum (6–8 weeks). Structural equation modeling (SEM) was used to examine associations between risk factors and latent factors: distress (composed of depression, generalized anxiety, irritability, and panic symptoms); fear (social anxiety, agoraphobia, specific phobia, and obsessive‐compulsive); and bipolar (mania and obsessive‐compulsive). Results: During pregnancy, past psychiatric history, anxiety sensitivity, maladaptive coping, and age were significant risk factors. In the postpartum, negative maternal attitudes and past psychiatric history were only risk factors for symptoms that composed distress. Experiential avoidance mediated the relation between maladaptive coping and symptoms that composed fear. Conclusion: It is important to assess for psychological risk factors starting in pregnancy. This study identified critical risk factors that are associated with the underlying commonality among perinatal mood and anxiety symptoms. Some of the risk factors as well as the mediator are malleable (negative maternal attitudes, experiential avoidance), creating new possibilities for prevention and treatment of perinatal mood and anxiety disorder symptoms. [ABSTRACT FROM AUTHOR]

Cross-cultural research in social and behavioral sciences has expanded hugely over the past 50 years, but progress is currently hampered by a lack of appreciation of the profoundly differing principles and goals of two distinct traditions. The first is the main variant of cross-cultural psychology (CCP), focusing on how culture shapes individual psychological functioning. The second was pioneered by Hofstede. It studies societal differences, and we name it "comparative culturology" (CC). We explain how these two paradigms differ. CCP is grounded in psychology and typically looks for unobservable individual-level constructs, which supposedly exist independently of their measurement, to provide understanding of individual differences as affected by culture. CC is an interdisciplinary field whose roots and impact span sociology, anthropology, political science, economics, management studies, psychology, and beyond. CC measures cultural dimensions as group-level constructs created by researchers, which are best understood as ecological manifolds : conglomerates of conceptually and statistically associated variables (not necessarily held together by a single underlying factor) that collectively explain national (and other group) differences. Given these paradigmatic distinctions, the two fields need not, and cannot, use the same validation methods. They should co-exist and collaborate based on mutual appreciation of their differences, without attempts by either field to impose its idiosyncrasies on the other. [ABSTRACT FROM AUTHOR]

Although modal shift from private motorized vehicles to public and active transport is regarded as crucially important for a sustainable society, such a shift has not been successful in Japan. One reason may be that measures have not satisfied the latent factors underlying people's transportation mode choices. Moreover, these factors may have changed during the COVID-19 pandemic. This study is aimed at ascertaining what workers have emphasized in transportation mode choice, and how those emphases have affected their actual mode choice. We also examined how changes in individuals' emphases during the pandemic have affected their actual mode choices. We used a four-time panel survey with 1377 respondents from Japan that asked participants about important items related to their transportation mode choice and activity patterns at four time points before and during the COVID-19 pandemic. After conducting a factor analysis to ascertain the latent factors behind mode choice, we applied a logistic regression model showing whether or not the respondents used each transportation mode at each time point. A conditional logistic regression model with individual fixed effect was applied to the relationship between an individual's change in emphasis and actual mode choice. Factor analysis results indicated four latent factors: usability and reliability, personal time–space, attitude toward the environment and health, and ease of use and convenience. A higher emphasis on personal time–space encouraged car use and discouraged public transport use at any time points. This trend became stronger during the pandemic. Changes in the emphasis level of this factor for an individual also affected the actual mode choice. A higher emphasis on usability and reliability also discouraged public transport use. Ease of use and convenience encouraged public transport use, and attitude toward the environment and health neither encouraged nor discouraged actual mode choice at any time point. Emphases related to transportation mode choice changed over time. However, changes in individuals' emphasis altered the actual mode choice at some time points but not at others. These results should be considered in transportation planning in order to prompt desired modal shifts by transportation users. • Latent factors of transportation mode choice have changed during COVID-19 era. • Greater emphasis on personal time-space discourages public transport use. • COVID-19 pandemic increased the importance of personal time-space in mode choice. • Emphasis on attention to the environment and health does not affect mode choice. [ABSTRACT FROM AUTHOR]

This article extends Fama and French (FF) models of observed factors by introducing latent factors (LFs) to further extract information from FF residual returns. A diagonally dominant (DD) rather than a diagonal or sparse matrix structure is adopted in this study to estimate remaining covariance between disturbance terms. Such an enhanced factor (EF) model provides a more comprehensive analysis for portfolio selection in high dimensions and also has certain advantages of estimation stability and computational efficiency. It is shown that the proposed EF–DD approach achieves overall better performance than competing models in terms of portfolio variance and the net Sharpe ratio. [ABSTRACT FROM AUTHOR]

Reductions in natural habitats urge that we better understand species' interconnection and how biological communities respond to environmental changes. However, ecological studies of species' interactions are limited by their geographic and taxonomic focus which can distort our understanding of interaction dynamics. We focus on bird–plant interactions that refer to situations of potential fruit consumption and seed dispersal. We develop an approach for predicting species' interactions that accounts for errors in the recorded interaction networks, addresses the geographic and taxonomic biases of existing studies, is based on latent factors to increase flexibility and borrow information across species, incorporates covariates in a flexible manner to inform the latent factors, and uses a meta-analysis dataset from 85 individual studies. We focus on interactions among 232 birds and 511 plants in the Atlantic Forest, and identify 5% of pairs of species with an unrecorded interaction, but posterior probability that the interaction is possible over 80%. Finally, we develop a permutation-based variable importance procedure for latent factor network models and identify that a bird's body mass and a plant's fruit diameter are important in driving the presence of species interactions, with a multiplicative relationship that exhibits both a thresholding and a matching behavior. for this article are available online. [ABSTRACT FROM AUTHOR]

E-scooters have emerged as a popular micromobility option for short trips, with many cities embracing shared e-scooters to enhance convenience for travelers and reduce reliance on automobiles. Despite their rising popularity, there is a lack of clear understanding of how user preferences and adoption practices vary by location. This study aims to explore user and non-user attitudes towards e-scooter use in diverse urban settings. A meta-analysis of data from three surveys (N = 1197) conducted in Washington, D.C., Miami, FL, and Los Angeles, CA, was performed to compare e-scooter users and non-user profiles, mode choice factors, and attitudes and preferences towards e-scooter use. Additionally, machine learning (ML) and SHAP (SHapley Additive exPlanations) analysis were utilized to identify influential factors in predicting e-scooter use in each city. The results reveal that the majority of e-scooter users are 25 to 39 of age, male, with higher income and a bachelor’s degree, and 92% possess a driver’s license. Significant differences in attitudes between e-scooter users and non-users highlight the complexity of perceptions towards e-scooter usage. The ML model indicates that employment status negatively impacts the prediction of e-scooter users, while factors such as living without a car and using non-motorized modes positively influence e-scooter use. Educational background is a significant e-scooter mode choice factor in Washington, D.C. and Miami, whereas attitudinal questions on car and technology usage are influential in Los Angeles. These findings provide valuable insights into the factors shaping e-scooter adoption, informing urban transportation planning and policymaking and enhancing understanding of shared micromobility and its impact on urban mobility.

Identifying the footprint of controlling factors on the spring discharge is challenging under the impacts of human activities and climate change. The challenge can be addressed by (1) exploring the temporal scaling features of spring discharge; and (2) disentangling and quantifying the effects of latent factors on the spring discharge. Taking an example of the Niangziguan Spring in China, the study explores the spring discharge over a 49-year (1959–2007) time window. The results show that the spring discharge is controlled by the positive feedback mechanism before the beginning of groundwater exploitation, while human activities cause variability in spring discharge. The effects of latent factors on the spring discharge characterize the obvious temporal scale-dependent. The quantitative analysis indicates that the local recharge of groundwater possesses the largest contribution (43%) to spring discharge on the temporal scale of 4–16 months. The western North Pacific monsoon (WNPM) and Indian Summer monsoon (ISM) cause the smallest effect (3%) on spring discharge. Human activities have become one of the most important factors (27%) in controlling spring discharge. The results are useful to predict or simulate the groundwater dynamic processes in the study reported here considering that spring discharge is a natural outlet of groundwater. [ABSTRACT FROM AUTHOR]

A collaborative recommender system based on a latent factor model has achieved significant success in the field of personalized recommender systems. However, the latent factor model suffers from sparsity problems. It is also limited in its ability to extract non-linear data features, resulting in poor recommendation performance. Inspired by the success of deep learning in different application areas, we incorporate deep learning into our proposed method to overcome the above problems. In this paper, we propose a dual deep learning and embedding-based latent factor model that considers dense user and item feature vectors. The model combines the existing deep learning and latent factor models to extract deep abstractions and non-linear feature representations of the data for rating prediction. The core idea is to map the dense user and item vectors generated by embedding techniques into dual, fully connected deep neural network architectures. In these two separate architectures, it learns the non-linear representation of the input data. The method then predicts the rating score by integrating the factors obtained from the two independent structures using the inner product. From the experimental result, we observe that the proposed model outperformed state-of-the-art existing models in real-world datasets (MovieLens ML-100K and ML-1M). [ABSTRACT FROM AUTHOR]

The conventional Cox proportional hazards (PH) model typically assumes fully observed predictors and constant regression coefficients. However, some predictors are latent variables, each of which must be characterized by multiple observed indicators from various perspectives. Moreover, the predictor effects may vary with time in practice. Accommodating such latent variables and identifying temporal covariate effects are frequently of primary interest. This study proposes a generalized structural equation model to investigate the temporal effects of observed and latent risk factors on the hazards of interest. The proposed model comprises a confirmatory factor analysis model as the measurement equation and a varying-coefficient PH model with observed and latent predictors as the structural equation. A hybrid procedure that combines the expectation-maximization (EM) algorithm and the corrected estimating equation approach is developed to estimate unknown parameters and coefficient functions. Simulation studies demonstrate the satisfactory performance of the proposed method. An application to a health survey study reveals insights into risk factors for elders' life expectancy. [ABSTRACT FROM AUTHOR]

Results of a study of an actual scientific and technical problem of determining the road safety level on road sections are presented in the paper. The aim of the study is to develop an express method for assessing road safety based on improving the method of the final accident rate by reducing the partial accident rate with the required accuracy of determining the safety level on public roads. The formalization of the relationship between the parameters of road traffic conditions has been carried out to study the level of safety on highways. Latent factors of the parameters of traffic conditions have been obtained on the basis of reduction of their aggregate by the method of principal components, taking into account the relationship of partial accident rates. This condition has allowed to develop analytical models for determining road safety for implementation in the practice of conducting a road safety audit. A method for identifying hazardous areas on highways has been proposed in the paper. It includes 11 parameters of road conditions out of 18 partial accident rates proposed in the prototype and regulatory documents of Ukraine. The adequacy of the model is determined by deviation values of the obtained data according to the prototype model from the values of the relative accident rate, which were 3.22 and 18.61, respectively. The proposed model reduces the deviation value by six times and affects the accuracy of determining the safety level. Experimental studies have been conducted on 79 km of highways (on 385 sectors with constant values of traffic conditions parameters). The results of a comparative analysis of the safety level for the prototype model and the developed model have been identical and shown a decrease in time and labor costs by 1.6 times without loss of accuracy. Thus, the developed methodology is recommended to be applied for determining the potential danger of road sections during an express analysis of the road safety or when conducting a safety audit at various stages of the road life cycle.

Objective: Alzheimer's disease (AD) is a growing challenge worldwide, which is why the search for early‐onset predictors must be focused as soon as possible. Longitudinal studies that investigate courses of neuropsychological and other variables screen for such predictors correlated to mild cognitive impairment (MCI). However, one often neglected issue in analyses of such studies is measurement invariance (MI), which is often assumed but not tested for. This study uses the absence of MI (non‐MI) and latent factor scores instead of composite variables to assess properties of cognitive domains, compensation mechanisms, and their predictability to establish a method for a more comprehensive understanding of pathological cognitive decline. Methods: An exploratory factor analysis (EFA) and a set of increasingly restricted confirmatory factor analyses (CFAs) were conducted to find latent factors, compared them with the composite approach, and to test for longitudinal (partial‐)MI in a neuropsychiatric test battery, consisting of 14 test variables. A total of 330 elderly (mean age: 73.78 ± 1.52 years at baseline) were analyzed two times (3 years apart). Results: EFA revealed a four‐factor model representing declarative memory, attention, working memory, and visual–spatial processing. Based on CFA, an accurate model was estimated across both measurement timepoints. Partial non‐MI was found for parameters such as loadings, test‐ and latent factor intercepts as well as latent factor variances. The latent factor approach was preferable to the composite approach. Conclusion: The overall assessment of non‐MI latent factors may pose a possible target for this field of research. Hence, the non‐MI of variances indicated variables that are especially suited for the prediction of pathological cognitive decline, while non‐MI of intercepts indicated general aging‐related decline. As a result, the sole assessment of MI may help distinguish pathological from normative aging processes and additionally may reveal compensatory neuropsychological mechanisms. [ABSTRACT FROM AUTHOR]

This study considers a time-varying coefficient additive hazards model with latent variables to examine potential observed and latent risk factors for survival of interest. The model consists of two parts: confirmatory factor analysis to measure each latent factor through multiple observable variables and a varying coefficient additive hazards model to examine the time-varying effects of the observed and latent risk factors on the hazard function. A hybrid estimation procedure that combines the expectation-maximum algorithm and corrected estimating equation method is developed to estimate the unknown parameters and nonparametric cumulative hazard functions. The consistency and asymptotic normality of the proposed estimators are established, and the pointwise confidence intervals and general confidence bands of the nonparametric functions are constructed accordingly. A satisfactory performance of the proposed method is demonstrated through simulation studies. An application to a study of chronic kidney disease for Chinese type 2 diabetes patients is presented. [ABSTRACT FROM AUTHOR]

This article tackles the problem of multilabel learning with missing labels. For this problem, it is widely accepted that label correlations can be used to recover the ground-truth label matrix. Most of the existing approaches impose the low-rank assumption on the observed label matrix to exploit label correlations by decomposing it into two matrices, which describe the latent factors of instances and labels, respectively. The quality of these latent factors highly influences the recovery of ground-truth labels and the construction of the multilabel classification model. In this article, we propose recovering the ground-truth label matrix by regularized matrix factorization. Specifically, the latent factors of instances are regularized by the local topological structure derived from the feature space, which can be further used to induce an effective multilabel model. Moreover, the latent factors of labels and the label correlations are mutually adapted via label manifold regularization. In this way, the recovery of the ground-truth label matrix and the construction of the multilabel classification model are optimized jointly and can benefit from the regularized matrix factorization. Extensive experimental studies show that the proposed approach significantly outperforms the state-of-the-art algorithms on both full-label and missing-label data. [ABSTRACT FROM AUTHOR]

Large-scale association analysis between multivariate responses and predictors is of great practical importance, as exemplified by modern business applications including social media marketing and crisis management. Despite the rapid methodological advances, how to obtain scalable estimators with free tuning of the regularization parameters remains unclear under general noise covariance structures. In this article, we develop a new methodology called sequential scaled sparse factor regression (SESS) based on a new viewpoint that the problem of recovering a jointly low-rank and sparse regression coefficient matrix can be decomposed into several univariate response sparse regressions through regular eigenvalue decomposition. It combines the strengths of sequential estimation and scaled sparse regression, thus sharing the scalability and the tuning free property for sparsity parameters inherited from the two approaches. The stepwise convex formulation, sequential factor regression framework, and tuning insensitiveness make SESS highly scalable for big data applications. Comprehensive theoretical justifications with new insights into high-dimensional multi-response regressions are also provided. We demonstrate the scalability and effectiveness of the proposed method by simulation studies and stock short interest data analysis. [ABSTRACT FROM AUTHOR]

Random effects models are popular statistical models for detecting and correcting spurious sample correlations due to hidden confounders in genome-wide gene expression data. In applications where some confounding factors are known, estimating simultaneously the contribution of known and latent variance components in random effects models is a challenge that has so far relied on numerical gradient-based optimizers to maximize the likelihood function. This is unsatisfactory because the resulting solution is poorly characterized and the efficiency of the method may be suboptimal. Here, we prove analytically that maximum-likelihood latent variables can always be chosen orthogonal to the known confounding factors, in other words, that maximum-likelihood latent variables explain sample covariances not already explained by known factors. Based on this result, we propose a restricted maximum-likelihood (REML) method that estimates the latent variables by maximizing the likelihood on the restricted subspace orthogonal to the known confounding factors and show that this reduces to probabilistic principal component analysis on that subspace. The method then estimates the variance–covariance parameters by maximizing the remaining terms in the likelihood function given the latent variables, using a newly derived analytic solution for this problem. Compared to gradient-based optimizers, our method attains greater or equal likelihood values, can be computed using standard matrix operations, results in latent factors that do not overlap with any known factors, and has a runtime reduced by several orders of magnitude. Hence, the REML method facilitates the application of random effects modeling strategies for learning latent variance components to much larger gene expression datasets than possible with current methods. [ABSTRACT FROM AUTHOR]

In this paper, we first revisit the Koenker and Bassett variational approach to (univariate) quantile regression, emphasizing its link with latent factor representations and correlation maximization problems. We then review the multivariate extension due to Carlier et al. (Ann Statist 44(3):1165–92, 2016,; J Multivariate Anal 161:96–102, 2017) which relates vector quantile regression to an optimal transport problem with mean independence constraints. We introduce an entropic regularization of this problem, implement a gradient descent numerical method and illustrate its feasibility on univariate and bivariate examples. [ABSTRACT FROM AUTHOR]

A collaborative recommender system based on a latent factor model has achieved significant success in the field of personalized recommender systems. However, the latent factor model suffers from sparsity problems. It is also limited in its ability to extract non-linear data features, resulting in poor recommendation performance. Inspired by the success of deep learning in different application areas, we incorporate deep learning into our proposed method to overcome the above problems. In this paper, we propose a dual deep learning and embedding-based latent factor model that considers dense user and item feature vectors. The model combines the existing deep learning and latent factor models to extract deep abstractions and non-linear feature representations of the data for rating prediction. The core idea is to map the dense user and item vectors generated by embedding techniques into dual, fully connected deep neural network architectures. In these two separate architectures, it learns the non-linear representation of the input data. The method then predicts the rating score by integrating the factors obtained from the two independent structures using the inner product. From the experimental result, we observe that the proposed model outperformed state-of-the-art existing models in real-world datasets (MovieLens ML-100K and ML-1M).

Factor analysis is a flexible technique for assessment of multivariate dependence and codependence. Besides being an exploratory tool used to reduce the dimensionality of multivariate data, it allows estimation of common factors that often have an interesting theoretical interpretation in real problems. However, standard factor analysis is only applicable when the variables are scaled, which is often inappropriate, for example, in data obtained from questionnaires in the field of psychology, where the variables are often categorical. In this framework, we propose a factor model for the analysis of multivariate ordered and non-ordered polychotomous data. The inference procedure is done under the Bayesian approach via Markov chain Monte Carlo methods. Two Monte Carlo simulation studies are presented to investigate the performance of this approach in terms of estimation bias, precision and assessment of the number of factors. We also illustrate the proposed method to analyze participants' responses to the Motivational State Questionnaire dataset, developed to study emotions in laboratory and field settings. [ABSTRACT FROM AUTHOR]

Principal component analysis (PCA) is a classical and ubiquitous method for reducing data dimensionality, but it is suboptimal for heterogeneous data that are increasingly common in modern applications. PCA treats all samples uniformly so degrades when the noise is heteroscedastic across samples, as occurs, e.g., when samples come from sources of heterogeneous quality. This paper develops a probabilistic PCA variant that estimates and accounts for this heterogeneity by incorporating it in the statistical model. Unlike in the homoscedastic setting, the resulting nonconvex optimization problem is not seemingly solved by singular value decomposition. This paper develops a heteroscedastic probabilistic PCA technique (HePPCAT) that uses efficient alternating maximization algorithms to jointly estimate both the underlying factors and the unknown noise variances. Simulation experiments illustrate the comparative speed of the algorithms, the benefit of accounting for heteroscedasticity, and the seemingly favorable optimization landscape of this problem. Real data experiments on environmental air quality data show that HePPCAT can give a better PCA estimate than techniques that do not account for heteroscedasticity. [ABSTRACT FROM AUTHOR]

Modeling species distributions over space and time is one of the major research topics in both ecology and conservation biology. Joint Species Distribution models (JSDMs) have recently been introduced as a tool to better model community data, by inferring a residual covariance matrix between species, after accounting for species' response to the environment. However, these models are computationally demanding, even when latent factors, a common tool for dimension reduction, are used. To address this issue, Taylor-Rodriguez et al. (2017) proposed to use a Dirichlet process, a Bayesian nonparametric prior, to further reduce model dimension by clustering species in the residual covariance matrix. Here, we built on this approach to include a prior knowledge on the potential number of clusters, and instead used a Pitman–Yor process to address some critical limitations of the Dirichlet process. We therefore propose a framework that includes prior knowledge in the residual covariance matrix, providing a tool to analyze clusters of species that share the same residual associations with respect to other species. We applied our methodology to a case study of plant communities in a protected area of the French Alps (the Bauges Regional Park), and demonstrated that our extensions improve dimension reduction and reveal additional information from the residual covariance matrix, notably showing how the estimated clusters are compatible with plant traits, endorsing their importance in shaping communities.

Asthma is a chronic respiratory disease with complex etiology. Adverse childhood experiences (ACEs) have been linked to asthma in adulthood. Underlying potential mechanisms for the ACE-asthma relationship include stress-induced inflammatory pathways and immune dysregulation. We conducted a cross-sectional secondary data analysis of the 2013 Alberta ACE Survey to explore the relationship between latent ACE factors and self-reported adult asthma. We evaluated the underlying correlation structure among eight different ACEs using exploratory factor analysis. We conducted a logistic regression model to evaluate whether ACE factors retained from the factor analysis predicted self-reported asthma in adulthood. Results were reported as odds ratios (ORs) with 95% confidence intervals (CIs). We analyzed ACE survey results from 1207 participants. Factor analysis yielded four ACE latent factors: factor 1/relational violence, factor 2/negative home environment, factor 3/illness at home, and factor 4/sexual abuse. Results of the logistic regression showed that experiencing sexual abuse (OR: 3.23; 95% CI: 1.89, 5.23), relational violence (OR: 1.99; 95% CI: 1.17, 3.38), and being exposed to a negative home environment (OR: 1.86; 95% CI: 1.03, 3.35) were predictive of a diagnosis of asthma in adulthood, whereas living in a household with someone experiencing illness did not show an effect (OR: 1.38; 95% CI: 0.75, 2.56). Factor analysis provides an effectual approach to understand the long-term impact of ACEs on respiratory health. Our findings have important implications to understand the developmental origins of asthma in adulthood and inform interventions aimed at reducing the lasting negative impact of childhood adversities on future respiratory health. [ABSTRACT FROM AUTHOR]

Moderation testing through latent factor models is relatively underutilized in hospitality and tourism research. The purpose of this research is to highlight the differences in the treatment of measurements of reflective constructs as composite indices versus latent factors in moderating effect tests in hospitality research. For this research, we build our primer on the investigation of the differences in customer satisfaction with the perceived entertainment experience at a hospitality/tourism attraction, contingent on customers' personality trait extraversion, borrowed from the Big-Five mini marker inventory. Our findings illustrate the consequences of the measurement conceptualization and the representation of constructs in statistical models with interaction effects. While using composites simplifies the estimation of the regression paths and provides a reasonable sense of the direction of the effect and its statistical significance, it is not always aligned with the theoretical and conceptual underpinning of the employed constructs. A statistical model with composites may underestimate an interaction effect, whereas a model with a dichotomized moderator may overestimate the interaction effect. The findings of this research draw the attention of the hospitality and tourism research community on different representations of reflective constructs in their measurement and statistical models. [ABSTRACT FROM AUTHOR]

Statistical methods relating tensor predictors to scalar outcomes in a regression model generally vectorize the tensor predictor and estimate the coefficients of its entries employing some form of regularization, use summaries of the tensor covariate, or use a low dimensional approximation of the coecient tensor. However, low rank approximations of the coefficient tensor can suer if the true rank is not small. We propose a tensor regression framework which assumes a soft version of the parallel factors (PARAFAC) approximation. In contrast to classic PARAFAC where each entry of the coefficient tensor is the sum of products of rowspeci contributions across the tensor modes, the soft tensor regression (Softer) framework allows the row-specic contributions to vary around an overall mean. We follow a Bayesian approach to inference, and show that softening the PARAFAC increases model flexibility, leads to improved estimation of coefficient tensors, more accurate identication of important predictor entries, and more precise predictions, even for a low approximation rank. From a theoretical perspective, we show that employing Softer leads to a weakly consistent posterior distribution of the coefficient tensor, irrespective of the true or approximation tensor rank, a result that is not true when employing the classic PARAFAC for tensor regression. In the context of our motivating application, we adapt Softer to symmetric and semi-symmetric tensor predictors and analyze the relationship between brain network characteristics and human traits. [ABSTRACT FROM AUTHOR]

Recommender systems have the effect of guiding users to interesting objects in a large space of possible options based on their preferences. However, providing accurate recommendations for new users, who do not have any records, is one of the most challenging problems in recommender systems. To retrieve sufficient information of the new users, existing solutions often require significant user effort to answer numerous questions. In this paper, we aim to propose a novel method that can build initial user profiles by requiring minimum user effort. In particular, non-negative matrix factorization is first adopted to extract only a few representative latent factors of the available items, which effectively reduces the problem dimensionality. A hybrid of personalized and non-personalized approach has then been proposed to iteratively collect users' feedback on the most important latent factors. In comparison with two state-of-the-art approaches, the proposed method requires less user effort while significantly improving the accuracy of the initial user profile.

Copyright of Journal of Computer Engineering & Applications is the property of Beijing Journal of Computer Engineering & Applications Journal Co Ltd. 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.)

Interpretability and stability are two important features that are desired in many contemporary big data applications arising in statistics, economics, and finance. While the former is enjoyed to some extent by many existing forecasting approaches, the latter in the sense of controlling the fraction of wrongly discovered features which can enhance greatly the interpretability is still largely underdeveloped. To this end, in this article, we exploit the general framework of model-X knockoffs introduced recently in Candès, Fan, Janson and Lv [(2018), "Panning for Gold: 'model X' Knockoffs for High Dimensional Controlled Variable Selection," Journal of the Royal Statistical Society, Series B, 80, 551–577], which is nonconventional for reproducible large-scale inference in that the framework is completely free of the use of p-values for significance testing, and suggest a new method of intertwined probabilistic factors decoupling (IPAD) for stable interpretable forecasting with knockoffs inference in high-dimensional models. The recipe of the method is constructing the knockoff variables by assuming a latent factor model that is exploited widely in economics and finance for the association structure of covariates. Our method and work are distinct from the existing literature in which we estimate the covariate distribution from data instead of assuming that it is known when constructing the knockoff variables, our procedure does not require any sample splitting, we provide theoretical justifications on the asymptotic false discovery rate control, and the theory for the power analysis is also established. Several simulation examples and the real data analysis further demonstrate that the newly suggested method has appealing finite-sample performance with desired interpretability and stability compared to some popularly used forecasting methods. for this article are available online. [ABSTRACT FROM AUTHOR]

The problem of predicting students' performance has been recently tackled by using matrix factorization, a popular method applied for collaborative filtering based recommender systems. This problem consists of predicting the unknown performance or score of a particular student for a task s/he did not complete or did not attend, according to the scores of the tasks s/he did complete and the scores of the colleagues who completed the task in question. The solving method considers matrix factorization and a gradient descent algorithm in order to build a prediction model that minimizes the error in the prediction of test data. However, we identified two key aspects that influence the accuracy of the prediction. On the one hand, the model involves a pair of important parameters: the learning rate and the regularization factor, for which there are no fixed values for any experimental case. On the other hand, the datasets are extracted from virtual classrooms on online campuses and have a number of implicit latent factors. The right figures are difficult to ascertain, as they depend on the nature of the dataset: subject, size, type of learning, academic environment, etc. This paper proposes some approaches to improve the prediction accuracy by optimizing the values of the latent factors, learning rate, and regularization factor. To this end, we apply optimization algorithms that cover a wide search space. The experimental results obtained from real-world datasets improved the prediction accuracy in the context of a thorough search for predefined values. Obtaining optimized values of these parameters allows us to apply them to further predictions for similar datasets. [ABSTRACT FROM AUTHOR]

Many cities are creating policies and programs aimed at expanding bicycling mode shares. Attitudes towards bicycling in combination with weather conditions, however, can strongly influence the decision to ride. In locations with wide annual seasonal variation, attitudes can radically alter demand on bicycle networks across the year. Though researchers have looked at weather impacts on bicycling, the link between attitudinal factors that might impact riding decisions and seasonal variation remains understudied. This paper investigates heterogeneous taste preferences about the inclination to bicycle for riders who ride only in the warm weather and those who ride all year long (including during severe winters). This research relies on survey data from Edmonton, Canada and presents results from a hybrid discrete choice model. After controlling for age, sex, education, income, the supply of bike lanes, and the latent variable "bicycling inclination," the results indicate that attitudes have a significantly positive impact on the decision to ride across seasons. The findings suggest that public education and season-specific training programs—particularly aimed at adults and women—have the potential to increase bicycling all year around. [ABSTRACT FROM AUTHOR]

Drawing on justice motive theory (Lerner et al. in Berkowitz and Walster (eds) Advances in experimental social psychology. Academic Press, New York, 1976), in the present work we examine belief in a just world as an underlying (latent) factor, reflected in general and personal beliefs. We first conducted a meta-analysis encompassing 51 articles, 76 samples, 19 countries and 23,900 participants, which revealed a positive association between the two forms of belief in a just world across studies (meta-analytic r =.52 [95% CI =.48,.56], p <.001) and within each of the individual samples. These results suggest that general and personal belief in a just world could be treated as joint indicators of a latent belief in a just world factor. In a second study, an online sample of participants (n = 311 participants, 54% women, Mage = 37.78 years) completed measures of general and personal belief in a just world, which we treated as indicators of latent belief in a just world. Measures of subjective well-being and psychological well-being were treated as indicators of latent well-being. Indicators loaded strongly on their respective latent factors, and the latent factors were strongly positively correlated (r =.74, p <.001). These results suggest that the link between belief in a just world and well-being may be even more substantial than has been previously recognized. In accordance with the original conceptualization of belief in a just world in justice motive theory, our results suggest benefits to examining the shared and unique aspects of general and personal beliefs in a just world. [ABSTRACT FROM AUTHOR]

• Study passenger evacuation behavior using online surveys and hybrid choice models. • Four models examine attributes, latent factors, interactions, and time perception. • Fear of flying, safety knowledge, and risk perception shape evacuation decisions. • Interactions illuminate emergency response intentions in scenarios. • Time perception attribute enhances models and informs accident research. In transport accident rescue operations, the primary focus is on saving lives. However, people's behavior and choices in such dangerous situations are unforeseeable. This paper employs a hybrid choice model to investigate how unobservable human heterogeneity impacts passengers' decisions in airplane emergency evacuations. Three primary latent factors – fear of flying, knowledge of air safety, and no perception of risk – were identified based on indicators from fifteen statements. Subsequently, the hybrid choice model was expanded to include a time perception variable and interaction effects. The findings revealed that compared to Chinese respondents, individuals from the USA exhibit lower fear of flying, more air safety knowledge, and a higher risk perception in flying. All else being equal, aviation professionals can accept longer distances to exits than other passengers, and individuals with higher fear of flying choose middle exits more frequently. In contrast, those with lower risk perception prefer exits at either end of the cabin. Interaction terms indicated that increased air safety knowledge and lower risk perception contribute to a greater acceptance of more distant exits. A lack of risk perception makes traveling alone a significant factor in choosing an exit other than the frontmost, and people tended to avoid following others in densely populated evacuation flows. Regarding time perception, participants perceived the frontmost exit as the quickest, followed by the backmost exit, while the middle exits were viewed as the slowest. This study provides valuable insights for exit choice behavior modeling and airplane evacuation enhancement. [ABSTRACT FROM AUTHOR]

Many modern big data applications feature large scale in both numbers of responses and predictors. Better statistical efficiency and scientific insights can be enabled by understanding the large-scale response-predictor association network structures via layers of sparse latent factors ranked by importance. Yet sparsity and orthogonality have been two largely incompatible goals. To accommodate both features, in this paper, we suggest the method of sparse orthogonal factor regression (SOFAR) via the sparse singular value decomposition with orthogonality constrained optimization to learn the underlying association networks, with broad applications to both unsupervised and supervised learning tasks, such as biclustering with sparse singular value decomposition, sparse principal component analysis, sparse factor analysis, and spare vector autoregression analysis. Exploiting the framework of convexity-assisted nonconvex optimization, we derive nonasymptotic error bounds for the suggested procedure characterizing the theoretical advantages. The statistical guarantees are powered by an efficient SOFAR algorithm with convergence property. Both computational and theoretical advantages of our procedure are demonstrated with several simulations and real data examples. [ABSTRACT FROM AUTHOR]

The main purpose of this study was to explore the latent relations of the selected morphometric, physiological and biochemical parameters. Thirty-six variables (12 morphometric, 9 physiological and 15 biochemical variables) were measured on 317 male-entities aged 17 - 35 y/o. The obtained data were analysed through the factor analysis of the first and second order. The statistical analyses were performed with the IBM SPSS Statistics software package, version 20. The factorization of the first order enabled extraction of 12 latent factors that explain 74.8% of the total variance, while the factorization of the second order enabled extraction of five latent components that explain 51.39% of the total variance. The final results of this study confirm the main hypothesis that there exist the numbers of latent variables that explain the latent structure of selected biometric measures. The nature of the extracted latent factors/components in both orders of factorization is relatively clear, understandable, and easy to interpret. The higher projections of the manifest biometric variables on the extracted latent factors of the first and second order were accordingly with the nature of the measured variables. The results of this research might be considered as one step more in the holistic approach to the biometric measures. [ABSTRACT FROM AUTHOR]