Abstract:
For pointwise estimation of anisotropic regression function under strong mixing condition, a linear wavelet estimator was constructed by wavelet method, and the convergence rate over l^p\left( 1 \leqslant p \lt \infty \right) pointwise error was discussed in B_\tilde p,q^\vec s\left( \mathbfR^d \right). The results of the research show that the conclusions under strong mixing conditions are closer to practical applications than the results when the samples are independent, and when the anisotropy index of space B_\tilde p,q^\vec s\left( \mathbfR^d \right) is equal, the conclusion is consistent with the result in the isotropic Besov space.