Journal of Applied Statistics最新文献

Practical scenarios often present instances where the types of responses are different between multi-source different datasets, reflecting distinct attributes or characteristics. In this paper, an integrative rank-based regression is proposed to facilitate information sharing among varied datasets with multi-type responses. Taking advantage of the rank-based regression, our proposed approach adeptly tackles differences in the magnitude of loss functions. In addition, it can robustly handle outliers and data contamination, and effectively mitigate model misspecification. Extensive numerical simulations demonstrate the superior and competitive performance of the proposed approach in model estimation and variable selection. Analysis of genetic data on HNSC and LUAD yields results with biological explanations and confirms its practical usefulness.

Several continuous distributions have been proposed recently to provide more flexibility in modeling lifetime data. Among these, the Slashed class of models, particularly the Slashed Lomax ( $\mathrm{SL}$ ) distribution, has gained special attention. This asymmetric model has positive support and it is notable for its stochastic representation and ability to fit heavy-tailed datasets. Despite the increasing number of new continuous models catering to specific samples, there have been few statistical tools introduced to evaluate their goodness-of-fits. To address this deficit, we employ the methodology outlined in J.M. Nicolas [Introduction aux statistiques de deuxième espèce: Applications des logs-moments et des logs-cumulants à l'analyse des lois d'images radar, TS, Trait. Signal 19 (2002), pp. 139-167] derived from the Mellin Transform (MT) to provide new goodness-of-fit measures for the $\mathrm{SL}$ distribution. These measures consider both qualitative and quantitative aspects. We derive the MT for the $\mathrm{SL}$ distribution, calculate the log-cumulants, and construct the log-cumulant diagram. Further, we introduce a test statistic using a combination of Hotelling's $T^2$ statistic and the multivariate Delta method to test hypotheses about the log-cumulants. We apply the new methodology to two real databases in the context of survival analysis to show its effectiveness in evaluating the fit criteria. We conduct bootstrap experiments to assess the power of the proposed test and to evaluate the performance of the estimators. The results revealed that the adjustment tools performed well and that the log-cumulant method proved to be an effective estimation criterion.

The present research deals with statistical inference for the expectation of a function of a random vector based on biased samples. After highlighting with the help of a motivating example the need for conducting this study, using the concept of multivariate weighted distributions, a consistent and asymptotically normally distributed estimator is proposed and utilized for developing statistical inference. A Monte Carlo study is carried out to examine the performance of the estimator proposed. Finally, the analysis of a real-world data set illustrates the benefits of using the proposed methods for statistical inference.

Observational data often exhibit clustered structure, which leads to inaccurate estimates of exposure effect if such structure is ignored. To overcome the challenges of modelling the complex confounder effects in clustered data, we propose a Bayesian doubly robust estimator of causal effects with random intercept BART to enhance the robustness against model misspecification. The proposed approach incorporates the uncertainty in the estimation of the propensity score, potential outcomes and the distribution of individual-level and cluster-level confounders into the exposure effect estimation, thereby improving the coverage probability of interval estimation. We evaluate the proposed method in the simulation study compared with frequentist doubly robust estimators with parametric and nonparametric multilevel modelling strategies. The proposed method is applied to estimate the effect of limited food access on the mortality of cardiovascular disease in the senior population.

Zero autocorrelation test statistics of the portmanteau type are studied under dependence. The asymptotic distribution of statistics formed with weighted averages of the autocorrelation and partial autocorrelation functions is theoretically obtained and its accuracy is then analyzed via simulation and in an empirical application. In the simulation study, we find that the proposed statistics provide test with sizes quite close to their nominal, intended sizes and with power functions which show high sensitivity to deviations from the null. It also reveals, for all the lags studied, that the tests are increasingly precise as the sample size increases. An application to financial time series modeling is given where the importance of using robust portmanteau statistics is illustrated. Specifically, we show that traditional tests incur in large deviations from their nominal size, whereas robust tests do not.

In this article, we consider statistical inference based on dependent competing risks data from Marshall-Olkin bivariate Weibull distribution. The maximum likelihood estimates of the unknown model parameters have been computed by using Newton-Raphson method under adaptive Type II progressive hybrid censoring with partially observed failure causes. Existence and uniqueness of maximum likelihood estimates are derived. Approximate confidence intervals have been constructed via the observed Fisher information matrix using asymptotic normality property of the maximum likelihood estimates. Bayes estimates and highest posterior density credible intervals have been calculated under gamma-Dirichlet prior distribution by using Markov chain Monte Carlo technique. Convergence of Markov chain Monte Carlo samples is tested. In addition, a Monte Carlo simulation is carried out to compare the effectiveness of the proposed methods. Further, three different optimality criteria have been taken into account to obtain the most effective censoring plans. From these simulation study results it has been concluded that Bayesian technique produces superior outcomes. Finally, a real-life data set has been analyzed to illustrate the operability and applicability of the proposed methods.

Survival predictions for patients are becoming increasingly important in clinical practice as they play a crucial role in aiding healthcare professionals to make more informed diagnoses and treatment decisions. The nested case-control designs have been extensively utilized as a cost-effective solution in many large cohort studies across epidemiology and other research fields. To achieve accurate survival predictions of individuals from nested case-control studies, we propose a semiparametric model averaging approach based on the partly linear additive proportional hazards structure to avoid the curse of dimensionality. The inverse probability weighting method is considered to estimate the parameters of submodels used in model averaging. We choose the weights by maximizing the pseudo-likelihood function constructed for the aggregated model and discuss the asymptotic optimality of selected weights. Simulation studies are conducted to assess the performance of our proposed model averaging method in the nested case-control study. Furthermore, we apply the proposed approach to real data to demonstrate its superiority.

The Extended Cox model provides an alternative to the proportional hazard Cox model for modelling data including time-varying covariates. Incorporating time-varying covariates is particularly beneficial when dealing with survival data, as it can improve the precision of survival function estimation. Deep learning methods, in particular, the Deep-pseudo survival neural network (DSNN) model have demonstrated a high potential for accurately predicting right-censored survival data when dealing with time-invariant variables. The DSNN's ability to discretise survival times makes it a natural choice for extending its application to scenarios involving time-varying covariates. This study adapts the DSNN to predict survival probabilities for data with time-varying covariates. To demonstrate this, we considered two scenarios: significant and non-significant time-varying covariates. For significant covariates, the Brier scores were below 0.25 at all considered specific time points, while, in the non-significant case, the Brier scores were above 0.25. The results illustrate that the DSNN performed comparably to the extended Cox, the Dynamic-DeepHit and mulitivariate joint models and on the simulated data. A real-world data application further confirms the predictive potential of the DSNN model in modelling survival data with time-varying covariates.

The latent class profile model (LCPM) is a widely used technique for identifying distinct subgroups within a sample based on observations' longitudinal responses to categorical items. This paper proposes an expanded version of LCPM by embedding time-specific structures. Such development allows analysts to investigate associations between latent class memberships and time-dependent predictors at specific time points. We suggest a simultaneous estimation of latent class measurement parameters via the expectation-maximization (EM) algorithm, which yields valid point and interval estimators of associations between latent class memberships and covariates. We illustrate the validity of our estimation strategy via numerical studies. In addition, we demonstrate the novelty of the proposed model by analyzing the head and neck cancer data set.

Control charts are sophisticated graphical tools used to detect and control aberrant variations. Different control schemes are designed to continuously monitor and improve the process stability and performance. This study proposes a bivariate exponentially weighted moving average chart for joint monitoring of the mean vector of Gumbel's bivariate geometric (GBG) data. The performance of the proposed chart is compared with Hotelling's $T^2$ chart. The results of the study indicated that the proposed control chart performs uniformly and substantially better than Hotelling's $T^2$ chart. In addition to two real-life examples, an example based on simulated data is also considered and compared to existing charts to verify the superiority of the proposed chart. Based on the comparisons, it turns out that the MEWMA (GBG) chart outperforms Hotelling's $T^2$ chart and individual EWMA control chart.