Automatic Smoothing of the Log Periodogram.

摘要

Consider an estimate of the log spectral density based on smoothing the log periodogram with a smoothing spline. This estimate is also a windowed estimate. An unbiased estimate R(lambda) of the expected integrated mean square error can be obtained as a function of the smoothing, or bandwidth parameter lambda. The smoothing parameter is then chosen as that lambda which minimizes R(lambda). The degree of the smoothing spline (equivalently, the shape parameter of the window) can also be chosen this way. Results of some Monte Carlo experiments illustrating the effectiveness of the method are given.