Bayes estimators of a multivariate generalized hyperbolic partial regression model

Document Type : Research Paper


1 Statistician at the Nineveh Agriculture Directorate, Mosul, Iraq

2 College of Administration and Economics, University of Baghdad, Baghdad, Iraq


The matrix-variate generalized hyperbolic distribution belongs to the family of heavy-tailed mixed probability distributions and is considered to be one of the continuous skewed probability distributions. This distribution has wide applications in the field of economics, especially in stock modeling. This paper includes estimation the parameters of the multivariate semi-parametric regression model represented by the multivariate partial linear regression model when the random error follows the matrix-variate generalized hyperbolic distribution, using the Bayesian method when non-informative prior information is available and under the assumption that the shape parameters and the skewness matrix are known. In addition, the bandwidth parameter is estimated by a suggested way based on the normal distribution rule and the proposed kernel function based on the mixed Gaussian kernel function and studying the findings on the generated data in a way suggested for the model, comparing the estimators depending on the criterion of the mean sum of squares error. The two researchers concluded that the proposed kernel function is better than the Gaussian kernel
function in estimate the parameters.


[1] A. S. AL-Mouel and A. J. Mohaisen, 2017 Study on Bayes semiparametric regression, Journal of Advances in Applied Mathematics, 4 (2017) 197-207.
[2] G. P. Box and G. C. Tiao, Bayesian inference in statistical analysis, Addison Wesley publishing company, Inc. London, U.K. (1973).
[3] M. P. Gallaugher and P. D. McNicholas, Three skewed matrix variate distributions, Statistics & Probability Letters, Elsevier, 145 (2019) 103-109.
[4] W. Hardle, Smoothing technique with implementation in S, Spring-Verlage , New York, (1993).
[5] R. Hogg, et al., Introduction to mathematical statistics (6th edition), Macmillan publishers company, New york,(2005).
[6] I. C. F. Ipsen and J. D. Lee, Determinant approximations, arXiv: 1105.0437v1, Mathmatics (N.A), (2011) 1-15.
[7] A. E. Koudou and C. Ley, Characterizations of GIG Laws: A survey complemented with two new result, Proba. Surv., 11 (2014) 161-176.
[8] N. Langrene and X. Warin, 2019” Fast and stable multivariate kernel density estimation by fast sum updating, Journal of Computational and Graphical Statistics, 28 (2019) 596-608.
[9] A. J. Lemonte and G. M. Cordeiro, The exponentiated inverse gaussian, Journal of Statistics and Probability letters, 81 (2011) 506-517.
[10] D. K. Nagar and A. K. Gupta, Extended variate gamma and beta function, ournal of Multivariate Analysis, 122 (2013) 53-69.
[11] S. J. Press, Subjective and objective Bayesian statistics principles, models and application, John Wiley & Sons, Canada, (2003).
[12] M. Przystalski, Estimation of the covariance matrix in multivariate partially linear models, Journal of Multivariate Analysis, 123 (2014) 380 -385.
[13] W. R. Schucany and J. Sommers, Improved of kernel type density estimators, JASA, 353 (1977) 420-423.
[14] B. W. Silverman, Density estimation for statistics and data analysis, Chapman and Hall, London, (1986).
[15] L. Thabane and S. Drekic, Hypothesis testing for the generalized multivariate modified Bessel model, Journal of Multivariate Analysis, (2003) 360-374.
[16] L. Thabane and M. S. Haq, On the matrix-variate generalized hyperbolic distribution and its Bayesian applications, Journal of Theoretical and Applied Statistical Science, 6 (2004) 511-526.
[17] J. You, et al, Statistical inference for multivariate partially linear regression models the Canadian journal of statistics, 1 (2013) 1-22.
Volume 12, Issue 2
November 2021
Pages 961-975
  • Receive Date: 14 February 2021
  • Revise Date: 21 March 2021
  • Accept Date: 25 May 2021