Robust Estimations for power Spectrum in ARMA(1,1) Model Simulation Study

Abstract

Simulation Study

Abstract :

Robust statistics Known as, Resistance to mistakes resulting of the deviation of Check hypotheses of statistical properties ( Adjacent Unbiased  , The Efficiency of data taken from a wide range of probability distributions follow a normal distribution or a mixture of other distributions with different standard deviations.

 power spectrum function lead to, President role in the analysis of Stationary random processes, organized according to time, may be discrete random variables or continuous. Measuring  its total capacity as frequency function.

Estimation methods Share with the concept of nonparametric in the absence of a model with clearly defined parameters (Free distribution) part with the distributed according to the Normal distribution, while the other part is unknown distribution, and thus it became the distribution of tainted its parameters is  unknown, so it can be considered the Robust Methods is the highest level in grades nonparametric methods which is will be based on the conversion calculable test to a Standard formula Conducted by the convergence operations.

The aim of the Search finding the best estimator of Power spectrum With the mixed ARMA (1,1) model for time series follow a Normal distribution. By Using Simulation experiments, on samples [n=50,100,150,200,250],and Different virtual values for ∅ و θ . It has been Showen  in tables 1,2,3 The Obvious difference between all the initial default values and generated values , Which may give results far from the real system results.