The Study of the Simulation of the Efficiency of Wavelet Estimation of Trend Functions under Long-term Dependence Errors

In this paper, we examine the estimation for trend functions in a time series model with Gaussian dependent residues with the aid of wavelet techniques. Using the simulations on the five different test functions and the process and taking into account the desired function, the factors affecting the error in our estimation have been discussed. The results show that the error rate of the wavelet method depends on the long-term dependence length. Finally, according to our simulations, the wavelet estimator method is compared with the so called classical methods of Kernel estimation and the results revealed that Wavelet estimations are more efficient.