Wavelet Thresholding in Partially Linear Models: A Computation and Simulation
Partially linear models have a linear part as in the linear regression and a non-linear part similar to that in the non-parametric regression. The estimates in Partially Linear Models have been studied previously using traditional smoothing methods such as smoothing spline, kernel and piecewise polynomial smoothers. In this paper, a wavelet thresholding method for estimating the corresponding parameters in Partially Linear Models is presented. Extensive simulation results shows that wavelet smoothing approach is comparable to traditional smoothing methods when their assumptions are satisfied. But wavelet smoothing is often superior when assumptions about the smoothness of the underlying function of non-parametric part are not satisfied. The computational complexity is linear on the sample size.
Qu, Leming. (2003). "Wavelet Thresholding in Partially Linear Models: A Computation and Simulation". Applied Stochastic Models in Business and Industry, 19(3), 221-230. https://doi.org/10.1002/asmb.499