M. Alonso-Pena, I. Gijbels, R. M. Crujeiras
Circular data are periodic observations that can be regarded as points on a unit circumference. These types of data are found in a many different fields such as psuchology, meteorology or biology, and can be found jointly with other real-valued variables. We propose a general approach that allows to estimate, nonparametrically, regression functions with a circular predictor and a general response, which can be either a continuous or a discrete variable. The estimation consists of hte maximization of a circular kernel weighted log-likelihood. The obtainment of the bias and variance of the estimators is also addressed, as well as the problemn of the selection of the smoothing parameter. The proposal will be illustrated with some real data applications.
Palabras clave: circular data, data-driven smoothing selection, kernel regression, local likelihood
Programado
GT18 Estadística no paramétrica III. Inferencia no paramétrica para la regresión
9 de junio de 2022 17:10
A15
