Search

The normal distribution is the most popular model in applications to real data. We propose a new extension of this distribution, called the Kummer beta normal distribution, which presents greater flexibility to model scenarios involving skewed data. The new probability density function can be represented as a linear combination of exponentiated normal pdfs. We also propose analytical expressions for some mathematical quantities: Ordinary and incomplete moments, mean deviations and order statistics. The estimation of parameters is approached by the method of maximum likelihood and Bayesian analysis. Likelihood ratio statistics and formal goodnessof-fit tests are used to compare the proposed distribution with some of its sub-models and non-nested models. A real data set is used to illustrate the importance of the proposed model.

The classical works in finance and insurance for modeling asset returns is the Gaussian model. However, when modeling complex random phenomena, more flexible distributions are needed which are beyond the normal distribution. This is because most of the financial and economic data are skewed and have ＂fat tails＂. Hence symmetric distributions like normal or others may not be good choices while modeling these kinds of data. Flexible distributions like skew normal distribution allow robust modeling of high-dimensional multimodal and asymmetric data. In this paper, we consider a very flexible financial model to construct comonotonic lower convex order bounds in approximating the distribution of the sums of dependent log skew normal random variables. The dependence structure of these random variables is based on a recently developed generalized multivariate skew normal distribution, known as unified skew normal distribution. The approximations are used to calculate the risk measure related to the distribution of terminal wealth. The accurateness of the approximation is investigated numerically. Results obtained from our methods are competitive with a more time consuming method known as Monte Carlo method.

We introduce the beta generalized normal distribution which is obtained by compounding the beta and generalized normal [Nadarajah, S., A generalized normal distribution, \emph{Journal of Applied Statistics}. 32, 685--694, 2005] distributions. The new model includes as sub-models the beta normal, beta Laplace, normal, and Laplace distributions. The shape of the new distribution is quite flexible, specially the skewness and the tail weights, due to two additional parameters. We obtain general expansions for the moments. The estimation of the parameters is investigated by maximum likelihood. We also proposed a random number generator for the new distribution. Actual synthetic aperture radar were analyzed and modeled after the new distribution. Results could outperform the $\mathcal{G}^0$, $\mathcal{K}$, and $\Gamma$ distributions in several scenarios., Comment: 17 pages, 2 tables, 6 figures

We introduce a novel framework for model-free variable selection with matrix-valued predictors. To test the importance of rows, columns, and submatrices of the predictor matrix in terms of predicti...

Comparing two population means of network data is of paramount importance in a wide range of scientific applications. Many existing network inference solutions focus on global testing of entire networks, without comparing individual network links. Besides, the observed data often take the form of vectors or matrices, and the problem is formulated as comparing two covariance or precision matrices under a normal or matrix normal distribution. Moreover, many tests suffer from a limited power under a small sample size. In this article, we tackle the problem of network comparison, both global and simultaneous inferences, when the data come in a different format, i.e., in the form of a collection of symmetric matrices, each of which encodes the network structure of an individual subject. Such data format commonly arises in applications such as brain connectivity analysis and clinical genomics. We no longer require the underlying data to follow a normal distribution, but instead impose some moment conditions that are easily satisfied for numerous types of network data. Furthermore, we propose a power enhancement procedure, and show that it can control the false discovery, while it has the potential to substantially enhance the power of the test. We investigate the efficacy of our testing procedure through both an asymptotic analysis and a simulation study under a finite sample size. We further illustrate our method with an example of brain structural connectivity analysis.

In matrix-valued datasets the sampled matrices often exhibit correlations among both their rows and their columns. A useful and parsimonious model of such dependence is the matrix normal model, in which the covariances among the elements of a random matrix are parameterized in terms of the Kronecker product of two covariance matrices, one representing row covariances and one representing column covariance. An appealing feature of such a matrix normal model is that the Kronecker covariance structure allows for standard likelihood inference even when only a very small number of data matrices is available. For instance, in some cases a likelihood ratio test of dependence may be performed with a sample size of one. However, more generally the sample size required to ensure boundedness of the matrix normal likelihood or the existence of a unique maximizer depends in a complicated way on the matrix dimensions. This motivates the study of how large a sample size is needed to ensure that maximum likelihood estimators exist, and exist uniquely with probability one. Our main result gives precise sample size thresholds in the paradigm where the number of rows and the number of columns of the data matrices differ by at most a factor of two. Our proof uses invariance properties that allow us to consider data matrices in canonical form, as obtained from the Kronecker canonical form for matrix pencils.

Along with advances in technology, matrix data, such as medical/industrial images, have emerged in many practical fields. These data usually have high dimensions and are not easy to cluster due to their intrinsic correlated structure among rows and columns. Most approaches convert matrix data to multi dimensional vectors and apply conventional clustering methods to them, and thus, suffer from an extreme high-dimensionality problem as well as a lack of interpretability of the correlated structure among row/column variables. Recently, a regularized model was proposed for clustering matrix-valued data by imposing a sparsity structure for the mean signal of each cluster. We extend their approach by regularizing further on the covariance to cope better with the curse of dimensionality for large size images. A penalized matrix normal mixture model with lasso-type penalty terms in both mean and covariance matrices is proposed, and then an expectation maximization algorithm is developed to estimate the parameters. The proposed method has the competence of both parsimonious modeling and reflecting the proper conditional correlation structure. The estimators are consistent, and their limiting distributions are derived. We applied the proposed method to simulated data as well as real datasets and measured its clustering performance with the clustering accuracy (ACC) and the adjusted rand index (ARI). The experiment results show that the proposed method performed better with higher ACC and ARI than those of conventional methods.

In this article, we investigate group differences in phthalate exposure profiles using NHANES data. Phthalates are a family of industrial chemicals used in plastics and as solvents. There is increasing evidence of adverse health effects of exposure to phthalates on reproduction and neuro-development, and concern about racial disparities in exposure. We would like to identify a single set of low-dimensional factors summarizing exposure to different chemicals, while allowing differences across groups. Improving on current multi-group additive factor models, we propose a class of Perturbed Factor Analysis (PFA) models that assume a common factor structure after perturbing the data via multiplication by a group-specific matrix. Bayesian inference algorithms are defined using a matrix normal hierarchical model for the perturbation matrices. The resulting model is just as flexible as current approaches in allowing arbitrarily large differences across groups but has substantial advantages that we illustrate in simulation studies. Applying PFA to NHANES data, we learn common factors summarizing exposures to phthalates, while showing clear differences across groups.

Due to its flexibility and good restoration performance, the Expected Patch Log Likelihood (EPLL) method has attracted extensive attention and has been further developed. However, the basic EPLL method is mainly applied for gray image restoration. For color image denoising with different channel noise levels, concatenating the RGB values into a vector and applying the basic EPLL directly can produce false colors and artifacts. In this paper, a Multi-Channel Expected Patch Log Likelihood (MC-EPLL) method is proposed for color image denoising with different channel noise levels. Considering the within and between channel correlation, the noise model of the concatenated vector of RGB channels can be constructed as a Matrix Normal Distribution. Under the KL divergence framework, the MC-EPLL model can be derived by combining the noise model and Gaussian Mixture Model (GMM) based patch prior. Based on the half quadratic splitting (HQS) strategy, the MC-EPLL model is decomposed into two sub-minimization problems and the closed-form solution of each sub-problem can be obtained. Experiments show the feasibility and superiority of the proposed MC-EPLL over the compared denoising methods.

While matrices are usually used as the basic data structure for experiments with repeated measurements or longitudinal data, testing methods for the equality of two matrix distributions have not been fully discussed in the literature. In this article, we propose three methods to test the equality of two matrix distributions: the likelihood ratio test, the Frobenius norm methods and triangle tests. We present a simulation to compare their performance under the matrix normal distribution. We apply the testing methods to compare the US economy, as measured by closing prices of five market indices, before and after the US stock market crash of 2008.

The distance-based regression model has many applications in analysis of multivariate response regression in various fields, such as ecology, genomics, genetics, human microbiomics, and neuroimaging. It yields a pseudo $F$ test statistic that assesses the relation between the distance (dissimilarity) of the subjects and the predictors of interest. Despite its popularity in recent decades, the statistical properties of the pseudo $F$ test statistic have not been revealed to our knowledge. This study derives the asymptotic properties of the pseudo $F$ test statistic using spectral decomposition under the matrix normal assumption, when the utilized dissimilarity measure is the Euclidean or Mahalanobis distance. The pseudo $F$ test statistic with the Euclidean distance has the same distribution as the quotient of two Chi-squared-type mixtures. The denominator and numerator of the quotient are approximated using a random variable of the form $\xi\chi^2_d+\eta$, and the approximate error bound is given. The pseudo $F$ test statistic with the Mahalanobis distance follows an $F$ distribution. In simulation studies, the approximated distribution well matched the “exact" distribution obtained by the permutation procedure. The obtained distribution was further validated on H1N1 influenza data, aging human brain data, and embryonic imprint data.

In this paper, the class of weakly $$ \psi $$ -uniformly convex functions is introduced and studied. Such functions are defined on a convex subcone D of a linear space V endowed with the structure of a normal decomposition system $$ (V,G,(\cdot )_\downarrow ) $$ . In particular, superquadratic functions and c-strongly convex functions are recovered as special cases. Some results in Abramovich et al. (J Math Anal Appl 327: 1444–1460, 2007), intended for multidimensional superquadratic functions defined on $${{\mathbb {R}}_{+}^{m}}$$ , are extended to the case of weakly $$ \psi $$ -uniformly convex functions. The concept of $$\psi $$ -almost superadditivity of functions is presented and applied. Interpretations of the obtained results are demonstrated for some matrix normal decomposition systems.

Finite mixtures of regressions with fixed covariates are a commonly used model-based clustering methodology to deal with regression data. However, they assume assignment independence, i.e., the allocation of data points to the clusters is made independently of the distribution of the covariates. To take into account the latter aspect, finite mixtures of regressions with random covariates, also known as cluster-weighted models (CWMs), have been proposed in the univariate and multivariate literature. In this paper, the CWM is extended to matrix data, e.g., those data where a set of variables are simultaneously observed at different time points or locations. Specifically, the cluster-specific marginal distribution of the covariates and the cluster-specific conditional distribution of the responses given the covariates are assumed to be matrix normal. Maximum likelihood parameter estimates are derived using an expectation-conditional maximization algorithm. Parameter recovery, classification assessment, and the capability of the Bayesian information criterion to detect the underlying groups are investigated using simulated data. Finally, two real data applications concerning educational indicators and the Italian non-life insurance market are presented.

Author(s): Zhang, Lu; Banerjee, Sudipto | Abstract: Multivariate spatially-oriented data sets are prevalent in the environmental and physical sciences.Scientists seek to jointly model multiple variables, each indexed by a spatial location, to capture anyunderlying spatial association for each variable and associations among the different dependent variables.Multivariate latent spatial process models have proved effective in driving statistical inference andrendering better predictive inference at arbitrary locations for the spatial process. High-dimensionalmultivariate spatial data, which is the theme of this article, refers to data sets where the number of spatiallocations and the number of spatially dependent variables is very large. The field has witnessed substantialdevelopments in scalable models for univariate spatial processes, but such methods for multivariate spatialprocesses, especially when the number of outcomes is moderately large, are limited in comparison. Here,we extend scalable modeling strategies for a single process to multivariate processes. We pursue Bayesianinference which is attractive for full uncertainty quantification of the latent spatial process. Our approachexploits distribution theory for the Matrix-Normal distribution, which we use to construct scalableversions of a hierarchical linear model of coregionalization (LMC) and spatial factor models that deliverinference over a high-dimensional parameter space including the latent spatial process. We illustrate thecomputational and inferential benefits of our algorithms over competing methods using simulation studiesand an analysis of a massive vegetation index dataset.

In this paper, we study the log-likelihood function and Maximum Likelihood Estimate (MLE) for the matrix normal model for both real and complex models. We describe the exact number of samples needed to achieve (almost surely) three conditions, namely a bounded log-likelihood function, existence of MLEs, and uniqueness of MLEs. As a consequence, we observe that almost sure boundedness of log-likelihood function guarantees almost sure existence of an MLE, thereby proving a conjecture of Drton, Kuriki and Hoff. The main tools we use are from the theory of quiver representations, in particular, results of Kac, King and Schofield on canonical decomposition and stability., 26 pages

Joint modeling of spatially-oriented dependent variables is commonplace in the environmental sciences, where scientists seek to estimate the relationships among a set of environmental outcomes accounting for dependence among these outcomes and the spatial dependence for each outcome. Such modeling is now sought for massive data sets with variables measured at a very large number of locations. Bayesian inference, while attractive for accommodating uncertainties through hierarchical structures, can become computationally onerous for modeling massive spatial data sets because of its reliance on iterative estimation algorithms. This manuscript develops a conjugate Bayesian framework for analyzing multivariate spatial data using analytically tractable posterior distributions that obviate iterative algorithms. We discuss differences between modeling the multivariate response itself as a spatial process and that of modeling a latent process in a hierarchical model. We illustrate the computational and inferential benefits of these models using simulation studies and analysis of a Vegetation Index data set with spatially dependent observations numbering in the millions., Comment: 22 pages, 12 figures

Comparative biologists are often interested in inferring covariation between multiple biological traits sampled across numerous related taxa. To properly study these relationships, we must control for the shared evolutionary history of the taxa to avoid spurious inference. Existing control techniques almost universally scale poorly as the number of taxa increases. An additional challenge arises as obtaining a full suite of measurements becomes increasingly difficult with increasing taxa. This typically necessitates data imputation or integration that further exacerbates scalability. We propose an inference technique that integrates out missing measurements analytically and scales linearly with the number of taxa by using a post-order traversal algorithm under a multivariate Brownian diffusion (MBD) model to characterize trait evolution. We further exploit this technique to extend the MBD model to account for sampling error or non-heritable residual variance. We test these methods to examine mammalian life history traits, prokaryotic genomic and phenotypic traits, and HIV infection traits. We find computational efficiency increases that top two orders-of-magnitude over current best practices. While we focus on the utility of this algorithm in phylogenetic comparative methods, our approach generalizes to solve long-standing challenges in computing the likelihood for matrix-normal and multivariate normal distributions with missing data at scale., Comment: 29 pages, 7 figures, 2 tables, 3 supplementary sections

Dramatic increases in the size and dimensionality of many modern datasets make crucial the need for sophisticated methods that can exploit inherent structure and handle missing values. In this article we derive an expectation–maximization (EM) algorithm for the matrix normal distribution, a distribution well-suited for naturally structured data such as spatio-temporal data. We review previously established maximum likelihood matrix normal estimates, and then consider the situation involving missing data. We apply our EM method in a simulation study exploring errors across different dimensions and proportions of missing data. We compare these errors to those from three alternative methods and show that our proposed EM method outperforms them in all scenarios. Finally, we implement the proposed EM method in a novel way on a satellite image dataset to investigate land-cover classification separability.

Studying crime trends and tendencies is an important problem that helps to identify socioeconomic patterns and relationships of crucial significance. Finite mixture models are famous for their flexibility in modeling heterogeneity in data. A novel approach designed for accounting for skewness in the distributions of matrix observations is proposed and applied to the United States crime data collected between 2000 and 2012 years. Then, the model is further extended by incorporating explanatory variables. A step-by-step model development demonstrates differences and improvements associated with every stage of the process. Results obtained by the final model are illustrated and thoroughly discussed. Multiple interesting conclusions have been drawn based on the developed model and obtained model-based clustering partition.

Summary Matrix-valued data, where the sampling unit is a matrix consisting of rows and columns of measurements, are emerging in numerous scientific and business applications. Matrix Gaussian graphical models are a useful tool to characterize the conditional dependence structure of rows and columns. We employ non-convex penalization to tackle the estimation of multiple graphs from matrix-valued data under a matrix normal distribution. We propose a highly efficient non-convex optimization algorithm that can scale up for graphs with hundreds of nodes. We establish the asymptotic properties of the estimator, which requires less stringent conditions and has a sharper probability error bound than existing results. We demonstrate the efficacy of our proposed method through both simulations and real functional magnetic resonance imaging analyses.

The joint distribution of standardized traces of $\frac{1}{n}XX'$ and of $\Big(\frac{1}{n}XX'\Big)^2$, where the matrix $X:p\times n$ follows a matrix normal distribution is proved asymptotically to be multivariate normal under condition $\frac{{n}}{p}\overset{n,p\rightarrow\infty}{\rightarrow}c>0$. Proof relies on calculations of asymptotic moments and cumulants obtained using a recursive formula derived in Pielaszkiewicz et al. (2015). The covariance matrix of the underlying vector is explicitely given as a function of $n$ and $p$.

We present a hierarchical decomposition scheme for computing the n-dimensional integral of multivariate normal probabilities that appear frequently in statistics. The scheme exploits the fact that ...

Recovering dynamic 3D structures from 2D image observations is highly under-constrained because of projection and missing data, motivating the use of strong priors to constrain shape deformation. In this paper, we empirically show that the spatiotemporal covariance of natural deformations is dominated by a Kronecker pattern. We demonstrate that this pattern arises as the limit of a spatiotemporal autoregressive process, and derive a Kronecker Markov Random Field as a prior distribution over dynamic structures. This distribution unifies shape and trajectory models of prior art and has the individual models as its marginals. The key assumption of the Kronecker MRF is that the spatiotemporal covariance is separable into the product of a temporal and a shape covariance, and can therefore be modeled using the matrix normal distribution. Analysis on motion capture data validates that this distribution is an accurate approximation with significantly fewer free parameters. Using the trace-norm, we present a convex method to estimate missing data from a single sequence when the marginal shape distribution is unknown. The Kronecker-Markov distribution, fit to a single sequence, outperforms state-of-the-art methods at inferring missing 3D data, and additionally provides covariance estimates of the uncertainty.

Consider the evaluation of model-based functions of cumulative distribution functions that are integrals. When the cumulative distribution function does not have a tractable form but simulation of the multivariate distribution is easily feasible, we can evaluate the integral via a Monte Carlo sample, replacing the model-based distribution function by the empirical distribution function. Given a simulation sample of size N, the naive method uses $$O(N^{2})$$ comparisons to compute the empirical distribution function at all N sample vectors. To obtain faster computational speed when N needs to be large to achieve a desired accuracy, we propose methods modified from the popular merge sort and quicksort algorithms that preserve their average $$O(N\log _{2}N)$$ complexity in the bivariate case. The modified merge sort algorithm can be extended to the computation of a d-dimensional empirical distribution function at the observed values with $$O(N\log _{2}^{d-1}N)$$ complexity. Simulation studies suggest that the proposed algorithms provide substantial time savings when N is large.

Recurrence relations for integrals that involve the density of multivariate normal distributions are developed. These recursions allow fast computation of the moments of folded and truncated multivariate normal distributions. Besides being numerically efficient, the proposed recursions also allow us to obtain explicit expressions of low-order moments of folded and truncated multivariate normal distributions. Supplementary material for this article is available online.

We construct a d-dimensional discrete multivariate distribution for which any proper subset of its components belongs to a specific family of distributions. However, the joint d-dimensional distribution fails to belong to that family and in other words, it is ‘inconsistent’ with the distribution of these subsets. We also address preservation of this ‘inconsistency’ property for the symmetric Binomial distribution, and some discrete distributions arising from the multivariate discrete normal distribution.