Metrika最新文献

In this paper, we propose a smoothed quantile regression estimator and variable selection procedure for partially linear varying coefficient models with nonignorable nonresponse. To avoid the computational problem caused by the non-smooth quantile loss function, we employ the kernel smoothing method. To address the identifiability issue, we use an instrument and estimate the parametric propensity function based on the generalized method of moments. Once the propensity is estimated, we construct the bias-corrected estimating equations utilizing the inverse probability weighting approach. Then, we apply the empirical likelihood method to obtain an unbiased estimator. The asymptotic properties of the proposed estimators are established for both the parametric and nonparametric parts. Meanwhile, variable selection is considered by using the SCAD penalty. The finite-sample performance of the estimators is studied through simulations, and a real-data example is also presented.

In this research, we present two-stage and purely sequential methodologies for estimating the scale parameter of the Moore and Bilikam family of lifetime distributions (see Moore and Bilikam in IEEE Trans Reliabil 27:64–67, 1978). We propose our methodologies under the minimum risk point estimation setup, whereby we consider the modified LINEX loss function plus non linear sampling cost. We study some interesting exact distributional properties associated with our stopping rules. We also present simulation analyses using Weibull distribution (special case of the Moore and Bilikam family) to check the performance of our two-stage and purely sequential procedures. Finally, we provide a real data set from COVID-19 and analyze it using the Weibull model in support of the practical utility of our proposed two-stage methodology.

The general minimum lower order confounding (GMC) is a criterion for selecting designs when the experimenter has prior information about the order of the importance of the factors. The paper considers the construction of (3^{n-m}) designs under the GMC criterion. Based on some theoretical results, it proves that some large GMC (3^{n-m}) designs can be obtained by combining some small resolution IV designs T. All the results for (4le #{T} le 20) are tabulated in the table, where (#) means the cardinality of a set.

In this work, the mean test is considered under the condition that the number of dimensions p is much larger than the sample size n when the covariance matrix is represented as a linear structure as possible. At first, the estimator of coefficients in the linear structures of the covariance matrix is constructed, and then an efficient covariance matrix estimator is naturally given. Next, a new test statistic similar to the classical Hotelling’s (T^2) test is proposed by replacing the sample covariance matrix with the given estimator of covariance matrix. Then the asymptotic normality of the estimator of coefficients and that of a new statistic for the mean test are separately obtained under some mild conditions. Simulation results show that the performance of the proposed test statistic is almost the same as the Hotelling’s (T^2) test statistic for which the covariance matrix is known. Our new test statistic can not only control reasonably the nominal level; it also gains greater empirical powers than competing tests. It is found that the power of mean test has great improvement when considering the structure information of the covariance matrix, especially for high-dimensional cases. Moreover, an example with real data is provided to show the application of our approach.

In the literature, the sharp positive upper mean-variance bounds on the expectations of order statistics based on independent identically distributed random variables with the decreasing and increasing failure rates, have been recently presented. In this paper we determine analogous evaluations in the dual cases when the parent distributions have monotone reversed failure rates.

We introduce finite mixtures of Ising models as a novel approach to study multivariate patterns of associations of binary variables. Our proposed models combine the strengths of Ising models and multivariate Bernoulli mixture models. We examine conditions required for the local identifiability of Ising mixture models, and develop a Bayesian framework for fitting them. Through simulation experiments and real data examples, we show that Ising mixture models lead to meaningful results for sparse binary contingency tables with imbalanced cell counts. The code necessary to replicate our empirical examples is available on GitHub: https://github.com/Epic19mz/BayesianIsingMixtures.

In this paper, we consider quantile regression estimation for linear models with covariate measurement errors and nonignorable missing responses. Firstly, the influence of measurement errors is eliminated through the bias-corrected quantile loss function. To handle the identifiability issue in the nonignorable missing, a nonresponse instrument is used. Then, based on the inverse probability weighting approach, we propose a weighted bias-corrected quantile loss function that can handle both nonignorable missingness and covariate measurement errors. Under certain regularity conditions, we establish the asymptotic properties of the proposed estimators. The finite sample performance of the proposed method is illustrated by Monte Carlo simulations and an empirical data analysis.

We deal with parameter estimation for linear parabolic second-order stochastic partial differential equations in two space dimensions driven by two types of Q-Wiener processes based on high frequency data with respect to time and space. We propose minimum contrast estimators of the coefficient parameters based on temporal and spatial increments, and provide adaptive estimators of the coefficient parameters based on approximate coordinate processes. We also give an example and simulation results of the proposed estimators.

A major challenge arising from data integration pertains to data heterogeneity in terms of study population, study design, or study coordination. Ignoring such heterogeneity in data analysis can lead to the biased estimation. In this paper, regression analysis of the proportional hazards model with heterogeneity parameter is studied. We combine the Model-X Knockoffs procedure with fused LASSO approach to control the false discovery rate in the variable selection and learn the integrative data analysis of partially heterogeneous subgroups when the outcome of interest is time to event. A regularized working partial likelihood function is established and a trick of reparameterization is developed for the numerical calculation of the proposed estimator. Simulation studies are conducted to assess the finite-sample performance of the proposed method. A data example from a clinical trial in primary biliary cirrhosis study is analyzed to demonstrate the application of our proposed method.

We consider the nonparametric regression and the classification problems for (psi )-weakly dependent processes. This weak dependence structure is more general than conditions such as, mixing, association(cdots ) A penalized estimation method for sparse deep neural networks is performed. In both nonparametric regression and binary classification problems, we establish oracle inequalities for the excess risk of the sparse-penalized deep neural networks estimators. Convergence rates of the excess risk of these estimators are also derived. The simulation results displayed show that, the proposed estimators can work well than the non penalized estimators, and that, there is a gain of using this estimator.