Subjects -> STATISTICS (Total: 130 journals)
 The end of the list has been reached or no journals were found for your choice.
Similar Journals
 Mathematical Methods of StatisticsJournal Prestige (SJR): 0.43 Number of Followers: 4      Hybrid journal (It can contain Open Access articles) ISSN (Print) 1934-8045 - ISSN (Online) 1066-5307 Published by Springer-Verlag  [2469 journals]
• Jensen’s Inequality Connected with a Double Random Good

Abstract: In this paper, we define a multiple random good of order $$2$$ denoted by $$X_{12}$$ whose possible values are of a monetary nature. A two-risky asset portfolio is a multiple random good of order $$2$$ . It is firstly possible to establish its expected return by using a linear and quadratic metric. We secondly establish the expected return on $$X_{12}$$ denoted by $$\mathbf{P}(X_{12})$$ by using a multilinear and quadratic metric. An extension of the notion of mathematical expectation of $$X_{12}$$ is carried out by using the notion of $$\alpha$$ -norm of an antisymmetric tensor of order $$2$$ . An extension of the notion of variance of $$X_{12}$$ denoted by $$\textrm{Var}(X_{12})$$ is shown by using the notion of $$\alpha$$ -norm of an antisymmetric tensor of order $$2$$ based on changes of origin. An extension of the notion of expected utility connected with $$X_{12}$$ is considered. An extension of Jensen’s inequality is shown as well. We focus on how the decision-maker maximizes the expected utility connected with multiple random goods of order $$2$$ being chosen by her under conditions of uncertainty and riskiness.
PubDate: 2022-06-01

• Bounds on the Expectations of $$\boldsymbol{L}$$ -Statistics Based on iid
Life Distributions

Abstract: We consider the order statistics based on independent identically distributed non-negative random variables. We determine sharp upper bounds on the expectations of arbitrary linear combinations of order statistics, expressed in the scale units being the $$p$$ th roots of $$p$$ th raw moments of original variables for various $$p\geq 1$$ . The bounds are more precisely described for the single order statistics and spacings. The lower bounds are concluded from the upper ones.
PubDate: 2022-06-01

Abstract: In a variety of applicative fields the level of information in random quantities is commonly measured by means of the Shannon Entropy. In particular, in reliability theory and survival analysis, time-dependent generalizations of this measure of uncertainty have been considered to dynamically describe changes in the degree of information over time. The Residual Entropy and the Residual Varentropy, for example, have been considered in the specialized literature to measure the information and its variability in residual lifetimes. In a similar way, one can consider dynamic measures of information for past lifetimes, i.e., for random lifetimes of items when one assumes that their failures occur before a fixed inspection time. This paper provides a study of the Past Varentropy, defined as the dynamic measure of variability of information for past lifetimes. From this study emerges the interest on a particular family of lifetimes distributions, whose members satisfy the property to be the only ones having constant Past Varentropy.
PubDate: 2022-06-01

• Matrix Variate Distribution Theory under Elliptical Models—V: The
Non-Central Wishart and Inverted Wishart Distributions

Abstract: The non-central Wishart and inverted Wishart distributions are studied in this work under elliptical models; some distributional results are based on some generalizations of the well-known Kummer relations, which leds us to determine that some moments have a polynomial representation. Then the non-central $$F$$ and ‘‘studentized Wishart’’ distributions are derived in a general setting. After some generalizations, including the so called non-central generalized inverted Wishart distribution, the classical results based on Gaussian models are derived here as corollaries.
PubDate: 2022-03-01
DOI: 10.3103/S1066530722010021

• D-Optimal Designs for the Mitscherlich Non-Linear Regression Function

Abstract: Mitscherlich’s function is a well-known three-parameter non-linear regression function that quantifies the relation between a stimulus or a time variable and a response. It has many applications, in particular in the field of measurement reliability. Optimal designs for estimation of this function have been constructed only for normally distributed responses with homoscedastic variances. In this paper we generalize this literature to D-optimal designs for discrete and continuous responses having their distribution function in the exponential family. We also demonstrate that our D-optimal designs can be identical to and different from optimal designs for variance weighted linear regression.
PubDate: 2022-03-01
DOI: 10.3103/S1066530722010033

• Bounds on the Expectations of $$\boldsymbol{L}$$ -statistics from Iid
Symmetric Populations in Various Scale Units

Abstract: We consider the order statistics $$X_{1:n},\ldots,X_{n:n}$$ based on independent identically symmetrically distributed random variables. We determine sharp upper bounds in the properly centered linear combinations of order statistics $$\sum_{i=1}^{n}c_{i}(X_{i:n}-\mu)$$ , where $$(c_{1},\ldots,c_{n})$$ is an arbitrary vector of coefficients from the $$n$$ -dimensional real space, and $$\mu$$ is the symmetry center of the parent distribution, in various scale units. The scale units are constructed on the basis of absolute central moments of the parent distribution of various orders. The bounds are specified for single order statistics. The lower bounds are immediately concluded from the upper ones.
PubDate: 2021-07-01
DOI: 10.3103/S1066530721030030

• A Necessary Bayesian Nonparametric Test for Assessing Multivariate
Normality

Abstract: A novel Bayesian nonparametric test for assessing multivariate normal models is presented. Although there are extensive frequentist and graphical methods for testing multivariate normality, it is challenging to find Bayesian counterparts. The approach considered in this paper is based on the Dirichlet process and the squared radii of observations. Specifically, the squared radii are employed to transform the $$m$$ -variate problem into a univariate problem by relying on the fact that if a random sample is coming from a multivariate normal distribution then the square radii follow a particular beta distribution. While the Dirichlet process is used as a prior on the distribution of the square radii, the concentration of the distribution of the Anderson–Darling distance between the posterior process and the beta distribution is compared to that between the prior process and beta distribution via a relative belief ratio. Key results of the approach are derived. The procedure is illustrated through several examples, in which it shows excellent performance.
PubDate: 2021-07-01
DOI: 10.3103/S1066530721030029

• Inferential Results for a New Inequality Curve

Abstract: We propose inferential results for a new integrated inequality curve, related to a new index of inequality and specifically designed for capturing significant shifts in the lower and upper tails of income distributions. In the last decades, indeed, substantial changes mainly occurred in the opposite sides of income distributions, raising serious concern to policy makers. These phenomena has been observed in countries like US, Germany, UK, and France. Properties of the index and curve have been investigated, and applications to real data disclosed a new way to look at inequality. First inferential results for the index have been published, as well. It seems natural, now, to be interested also in inferential results for the integrated curve. To fill this gap in the literature, we introduce two empirical estimators for the integrated curve, and show their asymptotical equivalence. Afterwards, we state their consistency. Finally, we prove the weak convergence in the space $$C[0,1]$$ of the corresponding empirical process to a Gaussian process, which is a linear transformation of a Brownian bridge. An analysis of real data from the Bank of Italy Survey of Income and Wealth is also presented, on the base of the obtained inferential results.
PubDate: 2021-01-01
DOI: 10.3103/S1066530721010026

• Local Dvoretzky–Kiefer–Wolfowitz Confidence Bands

Abstract: In this paper, we revisit the concentration inequalities for the supremum of the cumulative distribution function (CDF) of a real-valued continuous distribution as established by Dvoretzky, Kiefer, Wolfowitz and revisited later by Massart in in two seminal papers. We focus on the concentration of the local supremum over a sub-interval, rather than on the full domain. That is, denoting $$U$$ the CDF of the uniform distribution over $$[0,1]$$ and $$U_{n}$$ its empirical version built from $$n$$ samples, we study $$\mathbb{P}\Big{(}\sup_{u\in[\underline{u},\overline{u}]}U_{n}(u)-U(u)>\varepsilon\Big{)}$$ for different values of $$\underline{u},\overline{u}\in[0,1]$$ . Such local controls naturally appear for instance when studying estimation error of spectral risk-measures (such as the conditional value at risk), where $$[\underline{u},\overline{u}]$$ is typically $$[0,\alpha]$$ or $$[1-\alpha,1]$$ for a risk level $$\alpha$$ , after reshaping the CDF $$F$$ of the considered distribution into $$U$$ by the general inverse transform $$F^{-1}$$ . Extending a proof technique from Smirnov, we provide exact expressions of the local quantities $$\mathbb{P}\Big{(}\sup_{u\in[\underline{u},\overline{u}]}U_{n}(u)-U(u)>\varepsilon\Big{)}$$ and $$\mathbb{P}\Big{(}\sup_{u\in[\underline{u},\overline{u}]}U(u)-U_{n}(u)>\varepsilon\Big{)}$$ for each $$n,\varepsilon,\underline{u},\overline{u}$$ . Interestingly these quantities, seen as a function of $$\varepsilon$$ , can be easily inverted numerically into functions of the probability level $$\delta$$ . Although not explicit, they can be computed and tabulated. We plot such expressions and compare them to the classical bound $$\sqrt{\frac{\ln(1/\delta)}{2n}}$$ provided by Massart inequality. We then provide an application of such result to the control of generic functional of the CDF, motivated by the case of the conditional value at risk. Last, we extend the local concentration results holding individually for each $$n$$ to time-uniform concentration inequalities holding simultaneously for all $$n$$ , revisiting a reflection inequality by James, which is of independent interest for the study of sequential decision making strategies.
PubDate: 2021-01-01
DOI: 10.3103/S1066530721010038

• Applying the Solution for the First Multiplicity of Types Equation to
Calculate Exact Approximations of the Probability Distributions of
Statistical Values

Abstract: We consider here the use of the solution for the first multiplicity of types equation to compute exact probability distributions of statistical values and their exact approximations. We consider $${\Delta}$$ -exact distributions as their exact approximations; $${\Delta}$$ -exact distributions differ from exact distributions by no more than a predetermined, arbitrarily small value $${\Delta}$$ . It is shown that the basis for the exact distribution computing method is an enumeration of search area elements for solution of a linear first multiplicity of type equation composed of multiplicity type vectors. Each element represents here the number of occurrences for elements of a certain type (any sign of an alphabet) in the considered sample. It is shown simultaneously, that the method for restricting the search area for solution of the first multiplicity of type equation is applied for calculating exact approximation. We give an expression defining the algorithmic complexity of exact distributions calculated using the first multiplicity solution method which is finite and allows for each value of alphabet power to determine the maximum sample size for which exact distributions can be calculated by the first multiplicity solution method using limited computing power. To estimate the algorithmic complexity of computing the exact approximations, we used the expression obtained for the first time for the number of first multiplicity equation’s solutions with limitation on the values of coordinates of solution vectors. An expression determining algorithmic complexity for computing the exact approximations using the solution method for the first multiplicity equation with the constraint on the values of solution vector coordinates was obtained. The statistic value of maximal frequency is used as a parameter for restricting the solution vector coordinates, the probability of its excess is less than a pre-determined, arbitrarily small value $${\Delta}$$ . This permits to calculate the exact approximations of the distributions differing from their exact distribution values by no more than a chosen value $${\Delta}$$ . Results for calculating the maximum sample sizes for which exact approximations can be computed are given. It is shown that the algorithmic complexity of computing exact distributions by many orders of magnitude exceeds the complexity of computing their exact approximations. It is shown that application of the first multiplicity method for computing exact approximations allows increasing the volume of samples by a factor of two or more for equal values of the alphabet power as compared to computing exact distributions.
PubDate: 2020-10-01
DOI: 10.3103/S1066530720040031

• Selecting an Augmented Random Effects Model

Abstract: There are many collaborative studies where the data are discrepant while uncertainty estimates reported in each study cannot be relied upon. The classical commonly used random effects model explains this phenomenon by additional noise with a constant heterogeneity variance. This assumption may be inadequate especially when the smallest uncertainty values correspond to the cases which are most deviant from the bulk of data. An augmented random effects model for meta-analysis of such studies is offered. It proposes to think about the data as consisting of different classes with the same heterogeneity variance only within each cluster. The choice of the classes is to be made on the basis of the classical or restricted likelihood. We discuss the properties of the corresponding procedures which indicate the studies whose heterogeneity effect is to be enlarged. The conditions for the convergence of several iterative algorithms are given.
PubDate: 2020-10-01
DOI: 10.3103/S1066530720040043

• Censored Gamma Regression with Uncertain Censoring Status

Abstract: In this paper, we consider the problem of censored Gamma regression when the censoring status is missing at random. Three estimation methods are investigated. They consist in solving a censored maximum likelihood estimating equation where missing data are replaced by values adjusted using either regression calibration or multiple imputation or inverse probability weights. We show that the resulting estimates are consistent and asymptotically normal. Moreover, while asymptotic variances in missing data problems are generally estimated empirically (using Rubin’s rules for example), we propose closed-form consistent variances estimates based on explicit formulas for the asymptotic variances of the proposed estimates. A simulation study is conducted to assess finite-sample properties of the proposed parameters and asymptotic variances estimates.
PubDate: 2020-10-01
DOI: 10.3103/S106653072004002X

• On a Time Dependent Divergence Measure between Two Residual Lifetime
Distributions

Abstract: Recently, a time-dependent measure of divergence has been introduced by Mansourvar and Asadi (2020) to assess the discrepancy between the survival functions of two residual lifetime random variables. In this paper, we derive various time-dependent results on the proposed divergence measure in connection to other well-known measures in reliability engineering. The proposed criterion is also examined in mixture models and a general class of survival transformation models which results in some well-known models in the lifetime studies and survival analysis. In addition, the time-dependent measure is employed to evaluate the divergence between the lifetime distributions of $$k$$ -out-of- $$n$$ systems and also to assess the discrepancy between the distribution functions of the epoch times of a non-homogeneous Poisson process.
PubDate: 2020-07-01
DOI: 10.3103/S1066530720030023

• On Some Models of Ordered Random Variables and Characterizations of
Distributions

Abstract: The concept of extended neighboring order statistics introduced in Asadi et al. (2001) is a general model containing models of ordered random variables that are included in the generalized order statistics. This model also includes several models of ordered random variables that are not included in the generalized order statistics and is a helpful tool in unifying characterization results from several models of ordered random variables. In this paper, some general classes of distributions with many applications in reliability analysis and engineering, such as negative exponential, inverse exponential, Pareto, negative Pareto, inverse Pareto, power function, negative power, beta of the first kind, rectangular, Cauchy, Raleigh, Lomax, etc., have been characterized by using the regression of extended neighboring order statistics and decreasingly ordered random variables.
PubDate: 2020-07-01
DOI: 10.3103/S1066530720030035

• Optimal Rates for Nonparametric F-Score Binary Classification via
Post-Processing

Abstract: This work studies the problem of binary classification with the F-score as the performance measure. We propose a post-processing algorithm for this problem which fits a threshold for any score base classifier to yield high F-score. The post-processing step involves only unlabeled data and can be performed in logarithmic time. We derive a general finite sample post-processing bound for the proposed procedure and show that the procedure is minimax rate optimal, when the underlying distribution satisfies classical nonparametric assumptions. This result improves upon previously known rates for the F-score classification and bridges the gap between standard classification risk and the F-score. Finally, we discuss the generalization of this approach to the set-valued classification.
PubDate: 2020-04-01
DOI: 10.3103/S1066530720020027

• Adaptive Minimax Testing for Circular Convolution

Abstract: Given observations from a circular random variable contaminated by an additive measurement error, we consider the problem of minimax optimal goodness-of-fit testing in a non-asymptotic framework. We propose direct and indirect testing procedures using a projection approach. The structure of the optimal tests depends on regularity and ill-posedness parameters of the model, which are unknown in practice. Therefore, adaptive testing strategies that perform optimally over a wide range of regularity and ill-posedness classes simultaneously are investigated. Considering a multiple testing procedure, we obtain adaptive i.e. assumption-free procedures and analyse their performance. Compared with the non-adaptive tests, their radii of testing face a deterioration by a log-factor. We show that for testing of uniformity this loss is unavoidable by providing a lower bound. The results are illustrated considering Sobolev spaces and ordinary or super smooth error densities.
PubDate: 2020-04-01
DOI: 10.3103/S1066530720020039

• Optimal Adaptive Estimation on $${\mathbb{R}}$$ or $${\mathbb{R}}^{{+}}$$
of the Derivatives of a Density

Abstract: In this paper, we consider the problem of estimating the $$d$$ -th order derivative $$f^{(d)}$$ of a density $$f$$ , relying on a sample of $$n$$ i.i.d. observations $$X_{1},\dots,X_{n}$$ with density $$f$$ supported on $${\mathbb{R}}$$ or $${\mathbb{R}}^{+}$$ . We propose projection estimators defined in the orthonormal Hermite or Laguerre bases and study their integrated $${\mathbb{L}}^{2}$$ -risk. For the density $$f$$ belonging to regularity spaces and for a projection space chosen with adequate dimension, we obtain rates of convergence for our estimators, which are optimal in the minimax sense. The optimal choice of the projection space depends on unknown parameters, so a general data-driven procedure is proposed to reach the bias-variance compromise automatically. We discuss the assumptions and the estimator is compared to the one obtained by simply differentiating the density estimator. Simulations are finally performed. They illustrate the good performances of the procedure and provide numerical comparison of projection and kernel estimators
PubDate: 2020-01-01
DOI: 10.3103/S1066530720010020

• Multi-level Bayes and MAP Monotonicity Testing

Abstract: In this paper, we develop Bayes and maximum a posteriori probability (MAP) approaches to monotonicity testing. In order to simplify this problem, we consider a simple white Gaussian noise model and with the help of the Haar transform we reduce it to the equivalent problem of testing positivity of the Haar coefficients. This approach permits, in particular, to understand links between monotonicity testing and sparse vectors detection, to construct new tests, and to prove their optimality without supplementary assumptions. The main idea in our construction of multi-level tests is based on some invariance properties of specific probability distributions. Along with Bayes and MAP tests, we construct also adaptive multi-level tests that are free from the prior information about the sizes of non-monotonicity segments of the function.
PubDate: 2020-01-01
DOI: 10.3103/S1066530720010032

• Kernel Selection in Nonparametric Regression

Abstract: In the regression model $$Y=b(X)+\sigma(X)\varepsilon$$ , where $$X$$ has a density $$f$$ , this paper deals with an oracle inequality for an estimator of $$bf$$ , involving a kernel in the sense of Lerasle et al. [13], selected via the PCO method. In addition to the bandwidth selection for kernel-based estimators already studied in Lacour et al. [12] and Comte and Marie [3], the dimension selection for anisotropic projection estimators of $$f$$ and $$bf$$ is covered.
PubDate: 2020-01-01
DOI: 10.3103/S1066530720010044

• Bounding the Expectation of the Supremum of Empirical Processes Indexed by
Hölder Classes

Abstract: In this note, we provide upper bounds on the expectation of the supremum of empirical processes indexed by Hölder classes of any smoothness and for any distribution supported on a bounded set in $$\mathbb{R}^{d}$$ . These results can alternatively be seen as non-asymptotic risk bounds, when the unknown distribution is estimated by its empirical counterpart, based on $$n$$ independent observations, and the error of estimation is quantified by integral probability metrics (IPM). In particular, IPM indexed by Hölder classes are considered and the corresponding rates are derived. These results interpolate between two well-known extreme cases: the rate $$n^{-1/d}$$ corresponding to the Wassertein-1 distance (the least smooth case) and the fast rate $$n^{-1/2}$$ corresponding to very smooth functions (for instance, functions from a RKHS defined by a bounded kernel).
PubDate: 2020-01-01
DOI: 10.3103/S1066530720010056

JournalTOCs
School of Mathematical and Computer Sciences
Heriot-Watt University
Edinburgh, EH14 4AS, UK
Email: journaltocs@hw.ac.uk
Tel: +00 44 (0)131 4513762