Statistics Seminar
By Ali Sadeghkhani
Tuesday, September 24, 2024, at 3:00 pm
University of Windsor | In-person | Lambton Tower 9-118
Title: Multivariate Interval-Valued Models in Frequentist and Bayesian Schemes
Abstract: Interval-valued data, a form of symbolic data, has gained attention in recent years, particularly in privacy-preserving methods where individual-level data is concealed and reported as intervals. While most existing literature has focused on single-variable scenarios, with a few exceptions like Samdai et al. (2023) on bivariate cases, the multivariate domain remains largely unexplored. This seminar presents a novel extension of parameter estimation techniques to multivariate interval-valued data, offering the first derivation of maximum likelihood (ML) estimators and their asymptotic distributions in this context. Additionally, a theoretical Bayesian framework, previously applied only to univariate data, is extended to multivariate cases. The performance of these estimators is examined through Monte Carlo simulations, with real-world data used to demonstrate the practical utility of the models.
Counts toward seminar attendance for MSc and PhD students in Math & Stats