CFE-CMStatistics 2024: Start Registration
View Submission - CFECMStatistics2024
A0545
Title: A data-driven smoothing parameter for circular kernel density estimation Authors:  Jose Ameijeiras-Alonso - Universidade de Santiago de Compostela (Spain) [presenting]
Abstract: A novel data-driven smoothing parameter is introduced specifically designed for circular kernel density estimation. By adapting the well-established Sheather and Jones bandwidths to the circular domain, unknown parameters are replaced with estimates derived through plug-in techniques. Theoretical support is provided for the method, deriving the asymptotic mean squared error of the density estimator, its derivatives, and its functionals for circular data. A comprehensive simulation study demonstrates the superior performance of the proposed selectors compared to existing data-driven smoothing parameters. Additionally, the practical application of the plug-in rules is illustrated using real-world data on the timing of car accidents.