On a plug-in smoothing parameter for circular data
This talk presents a novel data-based smoothing parameter for circular kernel density estimation. We introduce a circular adaptation of the renowned Sheather and Jones bandwidths, replacing unknown quantities with suitable estimates using plug-in ideas. Theoretical support is provided. Additionally, we conduct a comprehensive simulation study to compare the performance of our proposed selectors with existing data-based smoothing parameters.
Keywords: Circular data; Directional Statistics Kernel Density Estimation Plug-in rule Sheather and Jones bandwidth