Statistics Seminar
By Ali Sadeghkhani
Thursday, September 26, 2024, at 3:00 pm
University of Windsor | In-person | Lambton Tower 9-118
Title: On Inference of Boxplot Symbolic Data: Applications in Climatology
Abstract: This study explores inference methods for boxplot-valued data in a multivariate framework, using both Bayesian and frequentist approaches. Boxplot-valued data, a type of symbolic data, captures variability and distributional characteristics in complex and big datasets. We propose new methodologies for parameter and density estimation and validate them through simulations, comparing Bayesian and frequentist estimators. Applied to climatological data from the Berkeley Earth Surface Temperature Study, our approach analyzes summer temperatures across European countries which aggregates 1.6 billion temperature reports, providing insights into climatic trends. The study also discusses the strengths and limitations of both inferential methods.
Counts toward seminar attendance for MSc and PhD students in Math & Stats