Behzad Mansouri

Detecting outliers in GARCH(p,q) models

ABSTRACT This article is concerned with the outliers in GARCH models. An iterative procedure is given for testing the presence of any type of the four common outliers. Since the distribution of test statistic cannot be obtained analytically, its distributional behavior is investigated via a simulation study. The simulation study is based on estimation of residuals standard deviation (σν), which are obtained using two methods, median absolute deviation method (MAD), and omit-one method. The proposed procedure is employed for testing the presence of outliers in weekly light oil price Indexes of Iran during 1997 to 2010.

Discrimination Measures for Fractional Integrated Models Based on Wavelets

Discrimination measures have been well developed for stationary time series. However in a large number of phenomena, long-term dependencies are involved. In this article, we are dealing with discrimination of fractional integrated models. Kullback–Leibler and Chernoffs discrimination measures are approximated, using the discrete wavelet transform (DWT) for discrimination of these time series classes. The simulation study indicates low misclassification rate, related to the approximations of Kullback–Leibler and Chernoff discrimination measures. Application to problem of classifying seismic data showed that our procedure performs as well as other procedures.

Wavelet analysis of uniformly time-modulated processes

ABSTRACT Uniformly time-modulated processes (UTMP) offer a simple, though convenient model for many non stationary time series. However, in many applications, dependency is considered as a disturbing factor. For the UTMP using Weierstrass representation of modulating function, we demonstrated that the correlation between coefficients in wavelet domain is a decreasing function of number of wavelet vanishing moments and tends to zero in progressive rate as we move from finer scales to coarser scales.

Discrimination and clustering of earthquakes and explosions based on NDWT

ABSTRACT This article is concerned with the development of discrimination procedure for seismic data based on feature extraction. These features are the maximum and minimum (maximin) values of Nondecimated Wavelets Transform (NDWT), which were extracted from the P component of the original earthquake and explosion traces. The method provides small error rates in both discrimination and clustering seismic wave datasets. It has been shown that our procedure performs as well as other procedures and in some cases better than other classifications methods. Also, the performance of our procedure is evaluated via a simulation study.

On Approximating Distribution of the Quadratic Discriminant Function

ABSTRACT The quadratic discriminant function (QDF) with known parameters has been represented in terms of a weighted sum of independent noncentral chi-square variables. To approximate the density function of the QDF as m-dimensional exponential family, its moments in each order have been calculated. This is done using the recursive formula for the moments via the Steins identity in the exponential family. We validate the performance of our method using simulation study and compare with other methods in the literature based on the real data. The finding results reveal better estimation of misclassification probabilities, and less computation time with our method.