Quantification of stylistic differences between musical artists is of academic interest to the music community and is also useful for other applications, such as music information retrieval and recommendation systems. Information about stylistic differences can be obtained by comparing the performances of different artists across common musical pieces. In this article we develop a statistical methodology for identifying and quantifying systematic stylistic differences among artists that are consistent across audio recordings of a common set of pieces, in terms of several musical features. Our focus is on a comparison of 10 different orchestras, based on data from audio recordings of the nine Beethoven symphonies. As generative or fully parametric models of raw audio data can be highly complex and more complex than necessary for our goal of identifying differences between orchestras, we propose to reduce the data from a set of audio recordings down to pairwise distances between orchestras, based on different musical characteristics of the recordings, such as tempo, dynamics and timbre. For each of these characteristics, we obtain multiple pairwise distance matrices, one for each movement of each symphony. We develop a hierarchical multidimensional scaling (HMDS) model to identify and quantify systematic differences between orchestras in terms of these three musical characteristics and interpret the results in the context of known qualitative information about the orchestras. This methodology is able to recover several expected systematic similarities between orchestras as well as to identify some more novel results. For example, we find that modern recordings exhibit a high degree of similarity to each other, as compared to older recordings.

Foreign public issuers (FPIs) are required by the Securities and Exchanges Commission (SEC) to file Form 20-F as comprehensive annual reports. In an effort to increase the usefulness of 20-Fs, the SEC recently enacted a regulation to accelerate the deadline of 20-F filing from six months to four months after the fiscal year-end. The rationale is that the shortened reporting lag would improve the informational relevance of 20-Fs. In this work we propose a jump additive model to evaluate the SEC’s rationale by investigating the relationship between the timeliness of 20-F filing and its decision usefulness using the market data. The proposed model extends the conventional additive models to allow possible discontinuities in the regression functions. We suggest a two-step jump-preserving estimation procedure and show that it is statistically consistent. By applying the procedure to the 20-F study, we find a moderate positive association between the magnitude of the market reaction and the filing timeliness when the acceleration is less than $17$ days. We also find that the market considers the filings significantly more informative when the acceleration is more than $18$ days and such reaction tapers off when the acceleration exceeds $40$ days.

This paper considers the problem of modeling a firm’s expected return as a nonlinear function of its observable characteristics. We investigate whether theoretically-motivated monotonicity constraints on characteristics and nonstationarity of the conditional expectation function provide statistical and economic benefit. We present an interpretable model that has similar out-of-sample performance to black-box machine learning methods. With this model, the data provide support for monotonicity and time variability of the conditional expectation function. Additionally, we develop an approach for characteristic selection using loss functions to summarize the posterior distribution. Standard unexplained volume, short-term reversal, size, and variants of momentum are found to be significant characteristics, and there is evidence this set changes over time.

Concentrations of nitrogen provide a critical metric for understanding ecosystem function and water quality in lakes. However, varying approaches for quantifying nitrogen concentrations may bias the comparison of water quality across lakes and regions. Different measurements of total nitrogen exist based on its composition (e.g., organic versus inorganic, dissolved versus particulate), which we refer to as nitrogen species. Fortunately, measurements of multiple nitrogen species are often collected and can, therefore, be leveraged together to inform our understanding of the controls on total nitrogen in lakes. We develop a multivariate hierarchical statistical model that fuses speciated nitrogen measurements, obtained across multiple methods of reporting, in order to improve our estimates of total nitrogen. The model accounts for lower detection limits and measurement error that vary across lake, species and observation. By modeling speciated nitrogen, as opposed to previous efforts that mostly consider only total nitrogen, we obtain more resolved inference with regard to differences in sources of nitrogen and their relationship with complex environmental drivers. We illustrate the inferential benefits of our model using speciated nitrogen data from the LAke GeOSpatial and temporal database (LAGOS).

Data integration methods that analyze multiple sources of data simultaneously can often provide more holistic insights than can separate inquiries of each data source. Motivated by the advantages of data integration in the era of “big data,” we investigate feature selection for high-dimensional multiview data with mixed data types (e.g., continuous, binary, count-valued). This heterogeneity of multiview data poses numerous challenges for existing feature selection methods. However, after critically examining these issues through empirical and theoretically-guided lenses, we develop a practical solution, the Block Randomized Adaptive Iterative Lasso (B-RAIL) which combines the strengths of the randomized Lasso, adaptive weighting schemes and stability selection. B-RAIL serves as a versatile data integration method for sparse regression and graph selection, and we demonstrate the effectiveness of B-RAIL through extensive simulations and a case study to infer the ovarian cancer gene regulatory network. In this case study, B-RAIL successfully identifies well-known biomarkers associated with ovarian cancer and hints at novel candidates for future ovarian cancer research.

Since 1973, the U.S. State Department has been using electronic record systems to preserve classified communications. Recently, approximately 1.9 million of these records from 1973–77 have been made available by the U.S. National Archives. While some of these communication streams have periods witnessing an acceleration in the rate of transmission, others do not show any notable patterns in communication intensity. Given the sheer volume of these communications, far greater than what had been available until now, scholars need automated statistical techniques to identify the communications that warrant closer study. We develop a statistical framework that can identify from a large corpus of documents a handful that historians would consider more interesting. Our approach brings together techniques from nonparametric signal estimation, statistical hypothesis testing and modern optimization methods—leading to a set of tools that help us identify and analyze various geometrical aspects of the communication streams. Dominant periods of heightened activities, as identified through these methods, correspond well with historical events recognized by standard reference works on the 1970s.

In various fields, data recorded continuously during a time interval and curve data, such as spectral data, become common. These kinds of data can be interpreted as functional data. In this paper we have studied binary classification from only positive and unlabeled functional data (PU classification for functional data). Our first contribution is to present a simple classification algorithm for this problem. The key feature of the algorithm is that it is not required an estimation of the unknown class prior (or the constant probability that a positive object is labeled). It is worth noting that the idea of our method can be applied to kernel linear discriminant analysis for general data. Our second contribution is to prove that, under mild regularity conditions similar to those in a supervised context, the proposed algorithm can achieve perfect asymptotic classification in the context of PU classification. In fact, we show that the proposed algorithm works well not only in numerical experiments but also for real data examples. Moreover, as an important practical application, we have used the proposed algorithm to identify handball players at risk for anterior cruciate ligament (ACL) injury based on ground reaction force data.

This article is motivated by the problem of studying the joint effect of different chemical exposures on human health outcomes. This is essentially a nonparametric regression problem, with interest being focused not on a black box for prediction but instead on selection of main effects and interactions. For interpretability we decompose the expected health outcome into a linear main effect, pairwise interactions and a nonlinear deviation. Our interest is in model selection for these different components, accounting for uncertainty and addressing nonidentifiability between the linear and nonparametric components of the semiparametric model. We propose a Bayesian approach to inference, placing variable selection priors on the different components, and developing a Markov chain Monte Carlo (MCMC) algorithm. A key component of our approach is the incorporation of a heredity constraint to only include interactions in the presence of main effects, effectively reducing dimensionality of the model search. We adapt a projection approach developed in the spatial statistics literature to enforce identifiability in modeling the nonparametric component using a Gaussian process. We also employ a dimension reduction strategy to sample the nonlinear random effects that aids the mixing of the MCMC algorithm. The proposed MixSelect framework is evaluated using a simulation study, and is illustrated using data from the National Health and Nutrition Examination Survey (NHANES). Code is available on GitHub.

Estimation of model parameters of computer simulators, also known as calibration, is an important topic in many engineering applications. In this paper we consider the calibration of computer model parameters with the help of engineering design knowledge. We introduce the concept of sensible (calibration) variables. Sensible variables are model parameters, which are sensitive in the engineering modeling, and whose optimal values differ from the engineering design values. We propose an effective calibration method to identify and to determine appropriate levels for the sensible variables with limited physical experimental data. The methodology is applied to a composite fuselage simulation problem.

An important problem in contemporary immunology studies based on single-cell protein expression data is to determine whether cellular expressions are remodeled postinfection by a pathogen. One natural approach for detecting such changes is to use nonparametric two-sample statistical tests. However, in single-cell studies direct application of these tests is often inadequate, because single-cell level expression data from processed uninfected populations often contain attributes of several latent subpopulations with highly heterogeneous characteristics. As a result, viruses often infect these different subpopulations at different rates, in which case the traditional nonparametric two-sample tests for checking similarity in distributions are no longer conservative. In this paper, we propose a new nonparametric method for Testing Remodeling under Heterogeneity (TRUH) that can accurately detect changes in the infected samples compared to possibly heterogeneous uninfected samples. Our testing framework is based on composite nulls and is designed to allow the null model to encompass the possibility that the infected samples, though unaltered by the virus, might be dominantly arising from underrepresented subpopulations in the baseline data. The TRUH statistic, which uses nearest neighbor projections of the infected samples into the baseline uninfected population, is calibrated using a novel bootstrap algorithm. We demonstrate the nonasymptotic performance of the test via simulation experiments and also derive the large sample limit of the test statistic which provides theoretical support toward consistent asymptotic calibration of the test. We use the TRUH statistic for studying remodeling in tonsillar T cells under different types of HIV infection and find that, unlike traditional tests which do not have any heterogeneity correction, TRUH based statistical inference conforms to the biologically validated immunological theories on HIV infection.

Indirect comparisons of treatment-specific outcomes across separate studies often inform decision making in the absence of head-to-head randomized comparisons. Differences in baseline characteristics between study populations may introduce confounding bias in such comparisons. Matching-adjusted indirect comparison (MAIC) (Pharmacoeconomics 28 (2010) 935–945) has been used to adjust for differences in observed baseline covariates when the individual patient-level data (IPD) are available for only one study and aggregate data (AGD) are available for the other study. The approach weights outcomes from the IPD using estimates of trial selection odds that balance baseline covariates between the IPD and AGD. With the increasing use of MAIC, there is a need for formal assessments of its statistical properties. In this paper we formulate identification assumptions for causal estimands that justify MAIC estimators. We then examine large sample properties and evaluate strategies for estimating standard errors without the full IPD from both studies. The finite-sample bias of MAIC and the performance of confidence intervals based on different standard error estimators are evaluated through simulations. The method is illustrated through an example comparing placebo arm and natural history outcomes in Duchenne muscular dystrophy.

Motivated by the analysis of accelerometer data taken across a population of individuals, we introduce a specific finite mixture of hidden Markov models with particular characteristics that adapt well to the specific nature of this type of longitudinal data. Our model allows for the computation of statistics that characterize the physical activity of a subject (e.g., the mean time spent at different activity levels and the probability of the transition between two activity levels) without specifying the activity levels in advance but by estimating them from the data. In addition, this approach allows the heterogeneity of the population to be taken into account and subpopulations with homogeneous physical activity behavior to be defined. We prove that, under mild assumptions, this model implies that the probability of misclassifying a subject decreases at an exponential decay with the length of its measurement sequence. Model identifiability is also investigated. We also report a comprehensive suite of numerical simulations to support our theoretical findings. The method is motivated by and applied to the Physical Activity and Transit Survey.

Tumor cell population consists of genetically heterogeneous subpopulations, known as subclones. Bulk sequencing data using high-throughput sequencing technology provide total and variant DNA and RNA read counts for many nucleotide loci as a mixture of signals from different subclones. We present RNDClone as a tool to deconvolute the mixture and reconstruct the subclones with distinct DNA genotypes and RNA expression profiles. In particular, we infer the number and population frequencies of subclones as well as subclonal copy numbers, variant allele numbers and gene expression levels by jointly modeling DNA and RNA read counts from the same tumor samples based on generalized latent factor models. Incorporating data at the RNA level provides new insights into intra-tumor heterogeneity in addition to the existing DNA-based inference. Performance of RNDClone is assessed using simulated and real-world datasets, including an analysis of three samples from a lung cancer patient in The Cancer Genome Atlas (TCGA). A potential fatal subclone is identified from the primary tumor which could explain the rapid prognosis and sudden death of the patient despite a promising diagnosis by conventional standards. The R package $\mathtt{RNDClone}$ is available in the Supplementary Material (Zhou et al. (2020)) and online at https://github.com/tianjianzhou/RNDClone.

The integration of mobile health (mHealth) devices into behavioral health research has fundamentally changed the way researchers and interventionalists are able to collect data as well as deploy and evaluate intervention strategies. In these studies, researchers often collect intensive longitudinal data (ILD) using ecological momentary assessment methods which aim to capture psychological, emotional and environmental factors that may relate to a behavioral outcome in near real time. In order to investigate ILD collected in a novel, smartphone-based smoking cessation study, we propose a Bayesian variable selection approach for time-varying effect models, designed to identify dynamic relations between potential risk factors and smoking behaviors in the critical moments around a quit attempt. We use parameter-expansion and data-augmentation techniques to efficiently explore how the underlying structure of these relations varies over time and across subjects. We achieve deeper insights into these relations by introducing nonparametric priors for regression coefficients that cluster similar effects for risk factors while simultaneously determining their inclusion. Results indicate that our approach is well positioned to help researchers effectively evaluate, design and deliver tailored intervention strategies in the critical moments surrounding a quit attempt.

Electronic health records (EHRs) are increasingly used for clinical and comparative effectiveness research but suffer from missing data. Motivated by health services research on diabetes care, we seek to increase the quality of EHRs by focusing on missing values of longitudinal glycosylated hemoglobin (A1c), a key risk factor for diabetes complications and adverse events. Under the framework of multiple imputation (MI), we propose an individualized Bayesian latent profiling approach to capture A1c measurement trajectories subject to missingness. The proposed method is applied to EHRs of adult patients with diabetes in a large academic Midwestern health system between 2003 and 2013 and had Medicare A and B coverage. We combine MI inferences to evaluate the association of A1c levels with the incidence of acute adverse health events and examine patient heterogeneity across identified patient profiles. We investigate different missingness mechanisms and perform imputation diagnostics. Our approach is computationally efficient and fits flexible models that provide useful clinical insights.

For most pathogens, testing procedures can be used to distinguish between different strains with which individuals are infected. Due to the growing availability of such data, multistrain models have increased in popularity over the past few years. Quantifying the interactions between different strains of a pathogen is crucial in order to obtain a more complete understanding of the transmission process, but statistical methods for this type of problem are still in the early stages of development. Motivated by this demand, we construct a stochastic epidemic model that incorporates additional strain information and propose a statistical algorithm for efficient inference. The model improves upon existing methods in the sense that it allows for both imperfect diagnostic test sensitivities and strain misclassification. Extensive simulation studies were conducted in order to assess the performance of our method, while the utility of the developed methodology is demonstrated on data obtained from a longitudinal study of Escherichia coli O157:H7 strains in feedlot cattle.

Humans are concurrently exposed to chemically, structurally and toxicologically diverse chemicals. A critical challenge for environmental epidemiology is to quantify the risk of adverse health outcomes resulting from exposures to such chemical mixtures and to identify which mixture constituents may be driving etiologic associations. A variety of statistical methods have been proposed to address these critical research questions. However, they generally rely solely on measured exposure and health data available within a specific study. Advancements in understanding of the role of mixtures on human health impacts may be better achieved through the utilization of external data and knowledge from multiple disciplines with innovative statistical tools. In this paper we develop new methods for health analyses that incorporate auxiliary information about the chemicals in a mixture, such as physicochemical, structural and/or toxicological data. We expect that the constituents identified using auxiliary information will be more biologically meaningful than those identified by methods that solely utilize observed correlations between measured exposure. We develop flexible Bayesian models by specifying prior distributions for the exposures and their effects that include auxiliary information and examine this idea over a spectrum of analyses from regression to factor analysis. The methods are applied to study the effects of volatile organic compounds on emergency room visits in Atlanta. We find that including cheminformatic information about the exposure variables improves prediction and provides a more interpretable model for emergency room visits for respiratory diseases.

Phylogenetics uses alignments of molecular sequence data to learn about evolutionary trees. Substitutions in sequences are modelled through a continuous-time Markov process, characterised by an instantaneous rate matrix, which standard models assume is time-reversible and stationary. These assumptions are biologically questionable and induce a likelihood function which is invariant to a tree’s root position. This hampers inference because a tree’s biological interpretation depends critically on where it is rooted. Relaxing both assumptions, we introduce a model whose likelihood can distinguish between rooted trees. The model is nonstationary with step changes in the instantaneous rate matrix at each speciation event. Exploiting recent theoretical work, each rate matrix belongs to a nonreversible family of Lie Markov models. These models are closed under matrix multiplication, so our extension offers the conceptually appealing property that a tree and all its subtrees could have arisen from the same family of nonstationary models.

We adopt a Bayesian approach, describe an MCMC algorithm for posterior inference and provide software. The biological insight that our model can provide is illustrated through an analysis in which nonreversible but stationary and nonstationary but reversible models cannot identify a plausible root.

We consider the problem of modeling and clustering heterogeneous event data arising from coupled conflict event and social media data sets. In this setting conflict events trigger responses on social media, and, at the same time, signals of grievance detected in social media may serve as leading indicators for subsequent conflict events. For this purpose we introduce the Hawkes Binomial Topic Model (HBTM) where marks, Tweets and conflict event descriptions are represented as bags of words following a Binomial distribution. When viewed as a branching process, the daughter event bag of words is generated by randomly turning on/off parent words through independent Bernoulli random variables. We then use expectation–maximization to estimate the model parameters and branching structure of the process. The inferred branching structure is then used for topic cascade detection, short-term forecasting, and investigating the causal dependence of grievance on social media and conflict events in recent elections in Nigeria and Kenya.

The problem of maximizing cell type discovery under budget constraints is a fundamental challenge for the collection and analysis of single-cell RNA-sequencing (scRNA-seq) data. In this paper we introduce a simple, computationally efficient and scalable Bayesian nonparametric sequential approach to optimize the budget allocation when designing a large-scale experiment for the collection of scRNA-seq data for the purpose of, but not limited to, creating cell atlases. Our approach relies on the following tools: (i) a hierarchical Pitman–Yor prior that recapitulates biological assumptions regarding cellular differentiation, and (ii) a Thompson sampling multiarmed bandit strategy that balances exploitation and exploration to prioritize experiments across a sequence of trials. Posterior inference is performed by using a sequential Monte Carlo approach which allows us to fully exploit the sequential nature of our species sampling problem. We empirically show that our approach outperforms state-of-the-art methods and achieves near-Oracle performance on simulated and scRNA-seq data alike. HPY-TS code is available at https://github.com/fedfer/HPYsinglecell.

The Mars rover Curiosity carries an instrument called ChemCam to determine the composition of the soil and rocks via laser-induced breakdown spectroscopy (LIBS). Los Alamos National Laboratory has developed a simulation capability that can predict spectra from ChemCam, but there are major-scale differences between the prediction and observation. This presents a challenge when using Bayesian model calibration to determine the unknown physical parameters that describe the LIBS observations. We present an analysis of LIBS data to support ChemCam based on including a structured discrepancy model in a Bayesian model-calibration scheme. This is both a novel application and an illustration of the importance of setting scientifically informed and constrained discrepancy models within Bayesian model calibration.

Obtaining an adequate description of the behaviour of narwhals in a pristine environment is important to understand natural behaviour as well as providing the means to determine potential changes in behaviour directly or indirectly caused by human activity. Based on $\text{Acousonde}^{\text{TM}}$ data from five narwhals in Scoresby Sound, this paper aims at modelling buzzing and calling rates of East Greenland narwhals as functions of time, space and, possibly, autoregressive memory. Both buzzing and calling are sounds produced by narwhals. Buzzing is a way for the whale to navigate and locate prey using echolocation, while calling is associated with social communication between whales. Logistic regression models without and with autoregressive components are compared based on AIC and comparatively assessed using diagnostics from point process theory. Adding an autoregressive component appears to improve the models, and further improvements for the buzzing model are made with a non-GLM extension. Effects of extrinsic covariates and memory are presented and interpreted. Buzzing occurs at deeper depths, and initiations of buzzes are separated by refractory periods. A possible feeding area is identified. Calling occurs closer to the surface, and, while the probability of calling in general is lower than buzzing, it is more likely that calls are clustered together rather than spread randomly.

Functional brain imaging through electroencephalography (EEG) relies upon the analysis and interpretation of high-dimensional, spatially organized time series. We propose to represent time-localized frequency domain characterizations of EEG data as region-referenced functional data. This representation is coupled with a hierarchical regression modeling approach to multivariate functional observations. Within this familiar setting we discuss how several prior models relate to structural assumptions about multivariate covariance operators. An overarching modeling framework, based on infinite factorial decompositions, is finally proposed to balance flexibility and efficiency in estimation. The motivating application stems from a study of implicit auditory learning, in which typically developing (TD) children, and children with autism spectrum disorder (ASD) were exposed to a continuous speech stream. Using the proposed model, we examine differential band power dynamics as brain function is interrogated throughout the duration of a computer-controlled experiment. Our work offers a novel look at previous findings in psychiatry and provides further insights into the understanding of ASD. Our approach to inference is fully Bayesian and implemented in a highly optimized Rcpp package.

Confounding by unmeasured spatial variables has received some attention in the spatial statistics and causal inference literatures, but concepts and approaches have remained largely separated. In this paper we aim to bridge these distinct strands of statistics by considering unmeasured spatial confounding within a causal inference framework and estimating effects using outcome regression tools popular within the spatial literature. First, we show how using spatially correlated random effects in the outcome model, an approach common among spatial statisticians, does not necessarily mitigate bias due to spatial confounding, a previously published but not universally known result. Motivated by the bias term of commonly-used estimators, we propose an affine estimator which addresses this deficiency. We discuss how unbiased estimation of causal parameters in the presence of unmeasured spatial confounding can only be achieved under an untestable set of assumptions which will often be application specific. We provide a set of assumptions which describe how the exposure and outcome of interest relate to the unmeasured variables, and we show that this set of assumptions is sufficient for identification of the causal effect based on the observed data when spatial dependencies can be represented by a ring graph. We implement our method using a fully Bayesian approach applicable to any type of outcome variable. This work is motivated by and used to estimate the effect of county-level limited access to supermarkets on the rate of cardiovascular disease deaths in the elderly across the whole continental United States. Even though standard approaches return null or protective effects, our approach uncovers evidence of unobserved spatial confounding and indicates that limited supermarket access has a harmful effect on cardiovascular mortality.

The problem of fitting a folded normal distribution by maximum likelihood has been described as ‘not straightforward’, and alternatives such as EM proposed. We suggest here that it is in fact straightforward to fit such a distribution by direct numerical maximization of the likelihood. We demonstrate this in an example. The relevant R code is included.