M. Alonso Pena, R. M. Crujeiras
We consider a regression problem where the response variable has a circular nature, such as angles, directions or other types of periodic observations. When considering the classical regression function, defined as the conditional expectation of the response given the value of the covariate, one finds that it is not always defined in the context of circular responses. In addition, the estimation of other conditional characteristics can lead to a better understanding of the relationship between the variables. In this work we consider the estimation of the conditional modes, instead of the conditional mean. The estimator, fully nonparametric, is based on kernel functions. The consistency of the estimator is formally proven and its properties are also investigated through a simulation study. Finally, an application to animal escapology is given, providing interesting results on the escape behavior of zebrafish when chased by a predator.
Keywords: circular data, modal regression, nonparametric methods
Scheduled
Ramiro Melendreras Award III
November 7, 2023 4:50 PM
CC4: Room 2