|
As a non-parametric regression method, local linear regression has been studied exten-sively and many results show that it is consistently superior to kernel regression estima-tor, such as Nadaraya-Watson estimator and Gasser-Muller estimator. The superiority of local linear regression is mainly reflected in two aspects:firstly, it has a faster rate of convergence than kernel regression estimator; secondly, it is free from Boundary Effect. The optimal and super-optimal convergence rate of local linear regression in a continuous-time environment is available in Shen and Sun [13], however, there is no research relating to the boundary effect of local linear regression in a continuous-time environment, which is our work in this paper.
|