Panpan Ba | Searcharxiv

arXiv2026

Uniform Strong Convergence of Kernel-Type Estimators for m-LENQD Sequences with Applications

In this paper, we firs tderiv eth eunifor mstron gconsistenc yan dth ecorrespondin gconvergence rates for kernel estimators of both the distribution function and the probability density function for sequences of m-linearly extended negative quadrant dependent (m-LENQD) random variables under mild regularity conditions. The obtained results significantl yimprov ean dexten dexisting ones in the literature. As their theoretical applications, we further establish the uniform strong consistency and convergence rates for estimators of the hazard rate function and the mode of the density function. Numerical simulations and a real data analysis are conducted to assess the finite- sample performance of the proposed estimators, confirmin gth easymptoti cpropertie sderive din the theoretical analysis.

stat.ME↗

arXiv2026

Optimal subsampling approach for Cox models under large-scale length-biased right-censored data

With the rapid advancement of data acquisition technologies, massive survival data have become increasingly prevalent, posing substantial computational challenges for the standard Cox proportional hazards model under complex censoring and truncation mechanisms. This paper addresses the analysis of large-scale length-biased right-censored data within the Cox model framework. We propose a novel subsampling approach to enhance computational efficiency for large samples and establish theoretically that the resulting subsample estimator provides an efficient approximation to the full-data estimator, while preserving consistency and asymptotic normality. Explicit optimal subsampling probabilities are derived under A- and L-optimality criteria. To facilitate practical implementation, we develop an efficient two-step estimation algorithm. Extensive numerical simulations and an empirical analysis of U.S. airline flight operation data evaluate the estimation performance and computational gains under various censoring rates and sample sizes. The results demonstrate that the proposed method substantially reduces computational cost while preserving statistical accuracy, offering a practical and scalable solution for analyzing large-scale length-biased right-censored survival data.

stat.ME↗