Estimate the location matrix of a multivariate semiparametric regression model when the random error follows a matrix--variate generalized hyperbolic distribution

Document Type : Research Paper


1 Statistician at the Nineveh Agriculture Directorate, Mosul, Iraq

2 Nineveh Agriculture Directorate, Mosul, Iraq.


    The matrix-variate generalized hyperbolic distribution is heavy-tailed mixed continuous skewed probability distribution. This distribution has multi applications in the field of economics, risk management, especially in stock modeling.
This paper includes the estimate of the location matrix  $\theta$  for the multivariate partial linear regression model, which is one of the multivariate semiparametric regression models when the random error follows a matrix-variate generalized hyperbolic distribution in the Bayesian technique depending on non-informative and informative prior information, estimating the location matrix under balanced and unbalanced loss function and the shape parameters ($\lambda ,\psi ,\nu $), skewness matrix ($\delta $), the scale matrix $(\Sigma)$ are known. In addition, estimation the smoothing parameter by a proposed method depending on the rule of thumb, the proposed kernel function depending on the mixed Gaussian kernel. the researchers concluded when non-informative and informative prior information is available that the posterior probability distribution for the location matrix  $\theta$ is a matrix-variate generalized hyperbolic distribution, through the experimental side, it was found that the proposed kernel function is overriding than the Gaussian kernel function in estimate the location matrix and under informative prior information.


[1] A. S. AL-Mouel and A. J. Mohaisen , Study on Bayes Semiparametric Regression , J. Adv. Appl. Math., 2 (4)(2017) 197-207.
[2] O. Barndorff-Nielsen, distribution and distributions on hyperbolae , Scand.J. Statist., 5(1978) 151-157.
[3] G. P. Box and G. C. Tiao , Bayesian Inference in Statistical Analysis , Addison Wesley publishing company, Inc. London, U.K., 1973.
[4] T. Choi, J. Lee and A. Roy , A note on the bayes factor in a semiparametric regression model , J. Multivar. Anal., 100(6)(2009) 1316-1327.
[5] P. J. Deshmaps , Bayesian estimation of generalized hyperbolic skewed student GARCH models , Comput. Stat. Data Anal., Elsevier, 56(11) (2012) 3035-3054.
[6] M. P.B. Gallaugher and P.D. McNicholas , Three Skewed Matrix Variate Distributions , Stat. Probab. Lett., Elsevier, 145 (2019) 103-109.
[7] W. Hardle, Smoothing Technique with Implementation in S, Spring-Verlage, New York., 1991.
[8] M. J. Jozani, E. Marchand and A. Parzian, Bayesian and robust Bayesian analysis under ageneral class of balanced loss function , Stat. Pap., 53 (2012) 51-60,
[9] A. E. Koudou and C. Ley , Characterizations of GIG Laws: A survey Complemented with Two New Result, Proba. Surv., 11 (2014) 161-176.
[10] N. Langrene and X. Warin, Fast and Stable Multivariate Kernel Density Estimation by Fast Sum Updating, J. Comput. Graphical Stat., 28 (2019) 596-608.
[11] A. J. Lemonte and G. M. Cordeiro , The exponentiated generalized inverse Gaussian distribution, J. Stat. Probab. lett., 81(4) (2011) .506-517,
[12] S. J. Press , Subjective and Objective Bayesian Statistics Principles, Models and Application, John Wiley & Sons, Canada. , 2003.
[13] M. Przystalski , Estimation of the Covariance Matrix in Multivariate Partially Linear Models, J. Multivar. Anal., 123 (2014) 380 -385.
[14] B. W. Silverman, Density Estimation for Statistics and Data Analysis, Chapman and Hall, London, 1986.
[15] L. Thabane and M. S. Haq , On the Matrix-variate Generalized Hyperbolic Distribution and its Bayesian Applications , J. Theor. . Appl. Stat. Sci. , 38(6) (2004) 511-526.
[16] J. You, Y. Zhou, G. Che , Statistical Inference for Multivariate Partially Linear Regression Models, Can. J. Stat.,41(1)(2013) 1-22.
Volume 13, Issue 1
March 2022
Pages 2467-2482
  • Receive Date: 19 June 2021
  • Revise Date: 11 October 2021
  • Accept Date: 03 November 2021