D. Bolón Rodríguez, R. M. Crujeiras Casais, A. Rodríguez Casal
The estimation of the highest density regions (HDR) of a statistical population is a useful tool in several fields. Some examples are the analysis of seismic data, the localization of minefields based on aerial observations and the detection of outliers within a sample. We propose a new non-parametric HDR estimator for directional data under some assumptions about the shape of the population HDR. Specifically, our estimation technique is a hybrid method that combines information of the kernel density estimator with smoothness assumptions on the class of sets considered.
Keywords: directional data, high density region estimation, non-parametric statistics
Scheduled
Nonparametric Statistics
June 10, 2022 4:00 PM
A13