On the some properties of a symmetric Grubbs' copula

L. K. Shiryaeva ; E. G. Repina

Journal of Samara State Technical University, Ser. Physical and Mathematical Sciences, Tome 22 (2018) no. 4, pp. 714-734.

Résumé

We investigate one-sided Grubbs’ statistics for a normal sample. Those statistics are standardized maximum and standardized minimum, i.e. studentized extreme deviation statistics. The two-parameter distribution of these statistics is considered, which arises when the one abnormal observation (outlier) differs from the other observations of its variance. We derive the formula for calculating the probability density function of studentized outlier deviation from sample average. A new two-parameter copula is extracted from the joint distribution of Grubbs’ statistics. The Grubbs’ copula is proved to be symmetric. As a result, one-sided Grubbs’ statistics have the property of exchangeability. Computer simulation of scatterplots from Grubbs’ copula is being performed. The scatterplot analysis shows that the Grubbs’ copula describes the negative statistical dependence. To study the effect of the copula’s parameters on the strength of this dependence, the estimation of the Kendall’s tau rank correlation coefficient is performed. The estimation algorithm uses computer simulation and it is realized in the R-package. We find that the copula’s parameters $n$ and $\nu>0$ have a multidirectional influence on the Kendall’s tau coefficient. If we do not change the parameter $\nu$ then the growth of the parameter $n$ leads to a decrease (in absolute value) of the Kendall’s tau coefficient, which reflects a decrease in the relationship’s strength between the marginals in Grubbs’ copula. If we do not change the parameter $n$, then growth of the parameter $\nu$ to 1 leads to a decrease in the Kendall’s tau coefficient (in absolute value), which reflects a decrease in the strength of the relationship. Further growth of the parameter $\nu$ leads to an increase in the Kendall’s tau coefficient (in absolute value), which reflects increased negative interdependence between the marginals.

Détail

Keywords: joint distribution function of maximum and standardized minimum, outlier, normal distribution, symmetric copula, exchangeability, Monte Carlo method
Mots-clés : Kendall's tau coefficient, R-code.

Bibliographie

[1] Blagoveshchenskii Yu. N., “Basics of copula's theory”, Applied Econometrics, 26:2 (2012), 113–130 (In Russian)

[2] Jwaid T., De Baets B., Kalicka J., Mesiar R., “Conic aggregation functions”, Fuzzy Sets and Systems, 167:1 (2011), 3–20 | DOI | Zbl

[3] Rodríguez-Lallena J. A., Úbeda-Flores M., “A new class of bivariate copulas”, Statistics and Probability Letters, 66:3 (2004), 315–325 | DOI | Zbl

[4] Kim J. M., Sungur E. A., Choi T., Heo T. Y., “Generalized bivariate copulas and their properties”, Model Assisted Statistics and Applications, 6:2 (2011), 127–136 | DOI

[5] Mesiar R., Najjari V., “New families of symmetric/asymmetric copulas”, Fuzzy Sets and Systems, 252:1 (2014), 99–110 | DOI | Zbl

[6] Nelsen R. B., An Introduction to Copulas, Lecture Notes in Statistics, Springer-Verlag, New York, 2006, xiii+269 pp. | DOI | Zbl

[7] Fantazzini D., “Analysis of Multidimensional Probability Distributions with Copula Functions, I”, Applied Econometrics, 22:2 (2011), 98–134 (In Russian)

[8] Ali Dolati, “On Dependence Properties of Random Minima and Maxima”, Communications in Statistics — Theory and Methods, 38:3 (2008), 393–399 | DOI

[9] Church C., The Asymmetric $t$-Copula with Individual Degrees of Freedom, A thesis submitted for the degree of MSc in Mathematical Finance, University of Oxford, Michaelmas, 2012, 48 pp. Available at (October 23, 2018) http://www.maths.ox.ac.uk/system/files/legacy/12804/The_Asymmetric_t-Copula_with_Individual_Degrees_of_Freedom.pdf

[10] Balaev A. I., “The copula based on multivariate $f$-distribution with vector of degrees of freedom”, Applied Econometrics, 33:1 (2014), 90–110 (In Russian)

[11] Shiryaeva L. K., “On tail dependence for Grubbs' copula-function”, Russian Math. (Iz. VUZ), 59:12 (2015), 56–72 | DOI | Zbl

[12] Kong-Sheng Zhang, Jin-Guan Lin, Pei-Rong Xu, “A new class of copulas involved geometric distribution: Estimation and applications”, Insurance: Mathematics and Economics, 66 (2016), 1–10 | DOI | Zbl

[13] Pearson E. S., Chandra Secar C., “The effciently of statistical tools and a criterion for the rejection of outlying observations”, Biometrika, 28:3–4 (1936), 308–320 | DOI | Zbl

[14] Grubbs F., “Sample Criteria for Testing Outlying observations”, Ann. Math. Statist., 21:1 (1950), 27–58 | DOI | Zbl

[15] Barnett V., Lewis T., Outliers in statistical data, John Wiley Sons, Chichester, 1984 | Zbl

[16] Zhang J., Keming Y., “The null distribution of the likelihood-ratio test for one or two outliers in a normal sample”, TEST, 15:1 (2006), 141–150 | DOI | Zbl

[17] Shiryaeva L. K., “On null and alternative distribution of statistics of two-side discordancy test for an extreme outlier”, Russian Math. (Iz. VUZ), 58:10 (2014), 52–66 | DOI | Zbl

[18] Shiryaeva L. K., “Calculation of power measures of Grabbs' criterion for checking for one outlier”, Sib. Zh. Ind. Mat., 13:4 (2010), 141–154 (In Russian) | MR | Zbl

[19] Shiryaeva L. K., “Using of special Hermite functions for investigation of power properties of Grubbs' criterion”, Vestn. Samar. Gos. Tekhn. Univ., Ser. Fiz.-Mat. Nauki [J. Samara State Tech. Univ., Ser. Phys. Math. Sci.], 2012, no. 4(29), 131–145 (In Russian) | DOI

[20] Shiryaeva L. K., “On distribution of Grubbs' statistics in case of normal sample with outlier”, Russian Math. (Iz. VUZ), 61:4 (2017), 72–88 | DOI | Zbl

[21] Jahnke E., Emde F., Lösch F., Tafeln höherer Funktionen [Tables of higher functions], B. G. Teubner Verlagsgesellschaft, Stuttgart, 1960, xii+318 pp. (In German) | Zbl | Zbl

[22] Galambos J., “Exchangeability”, Encyclopedia of statistical sciences, v. 2, eds. S. Kotz, and N. L. Johnson, Wiley, NY, 1986

[23] R Core Team., R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria, 2013 http://www.R-project.org/

[24] Ferguson S. T., Genest C., Hallin M., “Kendall’s tau for serial dependence”, Canadian Journal of Statistics, 28:3 (2000), 587–604 | DOI | Zbl

[25] Grothe O., Schnieders J., Sigers J., “Measuring association and dependence between random vectors”, Journal of Multivariate Analysis, 123 (2013), 96–110 | DOI

[26] Aivazian S. A., Mkhitaryan V. S., Applied Statistics and Essentials of Econometrics, v. 1, Probability Theory and Applied Statistics, Yuniti-Dana, Moscow, 2000 (In Russian)