Showing papers in "Journal of the American Statistical Association in 2019"
TL;DR: In this paper, a communication-efficient surrogate likelihood (CSL) framework for distributed statistical inference problems is presented, which provides a communication efficient surrogate to the global likelihoods.
Abstract: We present a communication-efficient surrogate likelihood (CSL) framework for solving distributed statistical inference problems. CSL provides a communication-efficient surrogate to the global like...
256 citations
TL;DR: A principled joint prior is developed for the range and the marginal variance of one-dimensional, two- dimensional, and three-dimensional Matérn GRFs with fixed smoothness and is applied to a dataset of annual precipitation in southern Norway, leading to conservative estimates of nonstationarity and improved predictive performance over the stationary model.
Abstract: Priors are important for achieving proper posteriors with physically meaningful covariance structures for Gaussian random fields (GRFs) since the likelihood typically only provides limited informat...
216 citations
TL;DR: A new forecast reconciliation approach is proposed that incorporates the information from a full covariance matrix of forecast errors in obtaining a set of coherent forecasts and minimizes the mean squared error of the coherent forecasts across the entire collection of time series under the assumption of unbiasedness.
Abstract: Large collections of time series often have aggregation constraints due to product or geographical groupings. The forecasts for the most disaggregated series are usually required to add-up exactly ...
208 citations
TL;DR: This work introduces a novel approach to Bayesian inference that improves robustness to small departures from the model: rather than conditioning on the event that the observed data are generated by the model, one conditions on theevent that the model generates data close to the observedData, in a distributional sense.
Abstract: The standard approach to Bayesian inference is based on the assumption that the distribution of the data belongs to the chosen model class. However, even a small violation of this assumption can ha...
183 citations
TL;DR: The main objective of the book is to provide a tool kit for practitioners to execute multiple imputation, and it avoids too much mathematical and technical details and uses graphical tools and visual displays to aid understanding.
Abstract: Missing data are frequently encountered in practice. A broader class of missing data is called incomplete data, which includes data with measurement error, multilevel data with latent variables, an...
183 citations
TL;DR: The deconfounder algorithm is developed, it is proved that it is unbiased, and it requires weaker assumptions than traditional causal inference, which is an effective approach to estimating causal effects in problems of multiple causal inference.
Abstract: Causal inference from observational data is a vital problem, but it comes with strong assumptions. Most methods assume that we observe all confounders, variables that affect both the causal...
167 citations
TL;DR: It is proved that the VB posterior converges to the Kullback–Leibler (KL) minimizer of a normal distribution, centered at the truth and the corresponding variational expectation of the parameter is consistent and asymptotically normal.
Abstract: A key challenge for modern Bayesian statistics is how to perform scalable inference of posterior distributions. To address this challenge, variational Bayes (VB) methods have emerged as a popular a...
167 citations
TL;DR: Subsampling Markov chain Monte Carlo is substantially more efficient than standard MCMC in terms of sampling efficiency for a given computational budget, and that it outperforms other subsampling methods for MCMC proposed in the literature.
Abstract: We propose subsampling Markov chain Monte Carlo (MCMC), an MCMC framework where the likelihood function for n observations is estimated from a random subset of m observations. We introduce a highly efficient unbiased estimator of the log-likelihood based on control variates, such that the computing cost is much smaller than that of the full log-likelihood in standard MCMC. The likelihood estimate is bias-corrected and used in two dependent pseudo-marginal algorithms to sample from a perturbed posterior, for which we derive the asymptotic error with respect to n and m, respectively. We propose a practical estimator of the error and show that the error is negligible even for a very small m in our applications. We demonstrate that subsampling MCMC is substantially more efficient than standard MCMC in terms of sampling efficiency for a given computational budget, and that it outperforms other subsampling methods for MCMC proposed in the literature. Supplementary materials for this article are availabl...
162 citations
TL;DR: In this article, the behavior of the Lasso for selecting invalid instruments in linear instrumental variables models for estimating causal effects of exposures on outcomes was investigated, as proposed recently by Ka...
Abstract: We investigate the behavior of the Lasso for selecting invalid instruments in linear instrumental variables models for estimating causal effects of exposures on outcomes, as proposed recently by Ka...
131 citations
TL;DR: In this article, the authors develop a methodology to conduct principal component analysis at high frequency and construct estimators of realized eigenvalues, eigenvectors, and principal components.
Abstract: We develop the necessary methodology to conduct principal component analysis at high frequency. We construct estimators of realized eigenvalues, eigenvectors, and principal components, and provide ...
131 citations
TL;DR: In this paper, a single spatial model that is able to capture both dependence classes in a parsimonious manner, and with a smooth transition between the two cases, is presented, covering a wide range of possibilities from asymptotic independence through to complete dependence.
Abstract: Many environmental processes exhibit weakening spatial dependence as events become more extreme. Well-known limiting models, such as max-stable or generalized Pareto processes, cannot capture this, which can lead to a preference for models that exhibit a property known as asymptotic independence. However, weakening dependence does not automatically imply asymptotic independence, and whether the process is truly asymptotically (in)dependent is usually far from clear. The distinction is key as it can have a large impact upon extrapolation, that is, the estimated probabilities of events more extreme than those observed. In this work, we present a single spatial model that is able to capture both dependence classes in a parsimonious manner, and with a smooth transition between the two cases. The model covers a wide range of possibilities from asymptotic independence through to complete dependence, and permits weakening dependence of extremes even under asymptotic dependence. Censored likelihood-based ...
TL;DR: In many branches of science, a large amount of data are being produced in many branches and many proven statistical methods are no longer applicable with extraordinary large datasets due to computational limitations as mentioned in this paper.
Abstract: Extraordinary amounts of data are being produced in many branches of science. Proven statistical methods are no longer applicable with extraordinary large datasets due to computational limitations....
TL;DR: It is argued that loss functions that are bounded, such as the classical biweight loss, are particularly suitable for changepoint detection—as it is shown that only bounded loss functions are robust to arbitrarily extreme outliers.
Abstract: Many traditional methods for identifying changepoints can struggle in the presence of outliers, or when the noise is heavy-tailed. Often they will infer additional changepoints to fit the outliers....
TL;DR: An extension of the glasso criterion (fglasso), which estimates the functional graphical model by imposing a block sparsity constraint on the precision matrix, via a group lasso penalty, and establishes the concentration inequalities of the estimates, which guarantee the desirable graph support recovery property.
Abstract: Graphical models have attracted increasing attention in recent years, especially in settings involving high-dimensional data. In particular, Gaussian graphical models are used to model the conditio...
TL;DR: Spatial regression models have been widely used to describe the relationship between a response variable and some explanatory variables over a region of interest, taking into account the spatial de... as mentioned in this paper.
Abstract: Spatial regression models have been widely used to describe the relationship between a response variable and some explanatory variables over a region of interest, taking into account the spatial de...
TL;DR: This article proposed incremental interventions that shift propensity score values rather than set treatments to fixed values, which avoid positivity assumptions entirely and allow longitudinal effects to be visualized with a single curve instead of lists of coefficients.
Abstract: Most work in causal inference considers deterministic interventions that set each unit’s treatment to some fixed value. However, under positivity violations these interventions can lead to nonidentification, inefficiency, and effects with little practical relevance. Further, corresponding effects in longitudinal studies are highly sensitive to the curse of dimensionality, resulting in widespread use of unrealistic parametric models. We propose a novel solution to these problems: incremental interventions that shift propensity score values rather than set treatments to fixed values. Incremental interventions have several crucial advantages. First, they avoid positivity assumptions entirely. Second, they require no parametric assumptions and yet still admit a simple characterization of longitudinal effects, independent of the number of timepoints. For example, they allow longitudinal effects to be visualized with a single curve instead of lists of coefficients. After characterizing incremental inter...
TL;DR: In this article, the causal interaction in factorial experiments is studied, in which several factors, each with multiple levels, are randomized to form a large number of possible treatment combinations.
Abstract: We study causal interaction in factorial experiments, in which several factors, each with multiple levels, are randomized to form a large number of possible treatment combinations. Examples of such...
TL;DR: In this article, the nonparametric least squares estimator (LSE) of a multivariate convex regression function was studied, given as the solution to a quadratic program with O(n 2 ) linear constraints.
Abstract: We study the nonparametric least squares estimator (LSE) of a multivariate convex regression function. The LSE, given as the solution to a quadratic program with O(n2) linear constraints (n being t...
TL;DR: The approach allows the estimation of a broad class of causal estimands and exact randomization-based p-values for testing causal effects, without imposing stringent assumptions, and derives a general central limit theorem that can be used to conduct conservative tests and build confidence intervals for causal effects.
Abstract: We define causal estimands for experiments on single time series, extending the potential outcome framework to dealing with temporal data. Our approach allows the estimation of a broad class of the...
TL;DR: The authors investigated the problem of inferring the causal predictors of a response from a set of explanatory variables (X1, X2, X3, X4, X5, X6).
Abstract: We investigate the problem of inferring the causal predictors of a response Y from a set of d explanatory variables (X1, …, Xd). Classical ordinary least-square regression includes all predictors t...
TL;DR: A generic nonparametric sure independence screening procedure, called BCor-SIS, is proposed on the basis of a recently developed universal dependence measure: Ball correlation, which shows strong screening consistency even when the dimensionality is an exponential order of the sample size without imposing sub-exponential moment assumptions on the data.
Abstract: Extracting important features from ultra-high dimensional data is one of the primary tasks in statistical learning, information theory, precision medicine, and biological discovery. Many of the sur...
TL;DR: In this paper, a class of partially linear functional additive models (PLFAM) is proposed to predict a scalar response by both parametric effects of a multivariate predictor and nonparametric effect of a multi-dimensional functional predictor.
Abstract: We investigate a class of partially linear functional additive models (PLFAM) that predicts a scalar response by both parametric effects of a multivariate predictor and nonparametric effects of a multivariate functional predictor. We jointly model multiple functional predictors that are cross-correlated using multivariate functional principal component analysis (mFPCA), and model the nonparametric effects of the principal component scores as additive components in the PLFAM. To address the high-dimensional nature of functional data, we let the number of mFPCA components diverge to infinity with the sample size, and adopt the component selection and smoothing operator (COSSO) penalty to select relevant components and regularize the fitting. A fundamental difference between our framework and the existing high-dimensional additive models is that the mFPCA scores are estimated with error, and the magnitude of measurement error increases with the order of mFPCA. We establish the asymptotic convergence ...
TL;DR: A robust multiple imputation-based approach to causal inference in this setting called PENCOMP is proposed, which builds on the penalized spline of propensity prediction method for missing data problems and tends to outperform doubly robust marginal structural modeling when the weights are variable.
Abstract: Valid causal inference from observational studies requires controlling for confounders. When time-dependent confounders are present that serve as mediators of treatment effects and affect future tr...
TL;DR: In this paper, the authors characterized node heterogeneity and link homophily in networks by node heterogeneity for which nodes exhibit different degrees of interaction and linkhomophily for nodes sharing common features tend to associate with eac...
Abstract: Networks are often characterized by node heterogeneity for which nodes exhibit different degrees of interaction and link homophily for which nodes sharing common features tend to associate with eac...
TL;DR: For multiple index models, it has been shown that sliced inverse regression (SIR) is consistent for estimating the sufficient dimension reduction (SDR) space if and only if ρ=limpn=0, where ρ is a constant.
Abstract: For multiple index models, it has recently been shown that the sliced inverse regression (SIR) is consistent for estimating the sufficient dimension reduction (SDR) space if and only if ρ=limpn=0 ,...
TL;DR: A fundamental problem in network analysis is clustering the nodes into groups which share a similar connectivity pattern as discussed by the authors, and existing algorithms for community detection assume the knowledge of the num....
Abstract: A fundamental problem in network analysis is clustering the nodes into groups which share a similar connectivity pattern. Existing algorithms for community detection assume the knowledge of the num...
TL;DR: This article derives theoretical results on the computational complexity of commonly used data augmentation algorithms and the Random Walk Metropolis algorithm for highly imbalanced binary data and shows that MCMC algorithms that exhibit a similar discrepancy will fail in large samples—a result with substantial practical impact.
Abstract: Many modern applications collect highly imbalanced categorical data, with some categories relatively rare. Bayesian hierarchical models combat data sparsity by borrowing information, while also qua...
TL;DR: In this article, treatment effect variation in randomized experiments has become essential for going beyond the "black box" of the average treatment effect and characterizing treatment effect variations in randomized trials.
Abstract: Understanding and characterizing treatment effect variation in randomized experiments has become essential for going beyond the “black box” of the average treatment effect. Nonetheless, traditional...
TL;DR: A new dynamic Tensor clustering method that works for a general-order dynamic tensor, and enjoys both strong statistical guarantee and high computational efficiency is proposed.
Abstract: Dynamic tensor data are becoming prevalent in numerous applications. Existing tensor clustering methods either fail to account for the dynamic nature of the data, or are inapplicable to a general-o...
TL;DR: In the last decade, significant theoretical advances have been made in the area of frequentist model averaging (FMA), however, the majority of this work has emphasized parametric model setups as mentioned in this paper.
Abstract: In the last decade, significant theoretical advances have been made in the area of frequentist model averaging (FMA); however, the majority of this work has emphasized parametric model setups. This...