The Annals of Statistics

Multivariate quantiles and multiple-output regression quantiles: From L1 optimization to halfspace depth

Marc Hallin, Davy Paindaveine, and Miroslav Šiman

Full-text: Open access

Abstract

A new multivariate concept of quantile, based on a directional version of Koenker and Bassett’s traditional regression quantiles, is introduced for multivariate location and multiple-output regression problems. In their empirical version, those quantiles can be computed efficiently via linear programming techniques. Consistency, Bahadur representation and asymptotic normality results are established. Most importantly, the contours generated by those quantiles are shown to coincide with the classical halfspace depth contours associated with the name of Tukey. This relation does not only allow for efficient depth contour computations by means of parametric linear programming, but also for transferring from the quantile to the depth universe such asymptotic results as Bahadur representations. Finally, linear programming duality opens the way to promising developments in depth-related multivariate rank-based inference.

Article information

Source
Ann. Statist. Volume 38, Number 2 (2010), 635-669.

Dates
First available: 19 February 2010

Permanent link to this document
http://projecteuclid.org/euclid.aos/1266586607

Digital Object Identifier
doi:10.1214/09-AOS723

Mathematical Reviews number (MathSciNet)
MR2604670

Subjects
Primary: 62H05: Characterization and structure theory
Secondary: 62J05: Linear regression

Keywords
Multivariate quantile quantile regression halfspace depth

Citation

Hallin, Marc; Paindaveine, Davy; Šiman, Miroslav. Multivariate quantiles and multiple-output regression quantiles: From L 1 optimization to halfspace depth. The Annals of Statistics 38 (2010), no. 2, 635--669. doi:10.1214/09-AOS723. http://projecteuclid.org/euclid.aos/1266586607.


Export citation

References

  • [1] Abdous, B. and Theodorescu, R. (1992). Note on the spatial quantile of a random vector. Statist. Probab. Lett. 13 333–336.
  • [2] Bassett, G. W., Koenker, R. and Kordas, G. (2004). Pessimistic portfolio allocation and Choquet expected utility. J. Fin. Econometrics 2 477–492.
  • [3] Breckling, J. and Chambers, R. (1988). M-quantiles. Biometrika 75 761–771.
  • [4] Breckling, J., Kokic, P. and Lübke, O. (2001). A note on multivariate M-quantiles. Statist. Probab. Lett. 55 39–44.
  • [5] Chakraborty, B. (2001). On affine equivariant multivariate quantiles. Ann. Inst. Statist. Math. 53 380–403.
  • [6] Chakraborty, B. (2003). On multivariate quantile regression. J. Statist. Plann. Inference 110 109–132.
  • [7] Chaudhuri, P. (1996). On a geometric notion of quantiles for multivariate data. J. Amer. Statist. Assoc. 91 862–872.
  • [8] Chen, W. W. and Deo, R. S. (2004). Power transformations to induce normality and their applications. J. R. Stat. Soc. Ser. B Stat. Methodol. 66 117–130.
  • [9] Donoho, D. L. and Gasko, M. (1992). Breakdown properties of location estimates based on halfspace depth and projected outlyingness. Ann. Statist. 20 1803–1827.
  • [10] Dudley, R. M. and Koltchinskii, V. I. (1992). The spatial quantiles. Unpublished manuscript.
  • [11] Fukuda, K. and Rosta, V. (2005). Data depth and maximum feasible subsystems. In Graph Theory and Combinatorial Optimization (D. Avis, A. Hertz and O. Marcotte, eds.) 37–67. Springer, New York.
  • [12] Gutenbrunner, C. and Jurečková, J. (1992). Regression rank scores and regression quantiles. Ann. Statist. 20 305–330.
  • [13] Haberman, S. J. (1989). Concavity and estimation. Ann. Statist. 17 1631–1661.
  • [14] Heinz, G., Peterson, L. J., Johnson, R. W. and Kerk, C. J. (2003). Exploring relationships in body dimensions. J. Statist. Education 11. Available at http://www.amstat.org/publications/jse/v11n2/datasets.heinz.html.
  • [15] Hettmansperger, T. P., Nyblom, J. and Oja, H. (1992). On multivariate notions of sign and rank. In L1-Statistical Analysis and Related Methods (Neuchâtel, 1992) 267–278. North-Holland, Amsterdam.
  • [16] Koenker, R. (2005). Quantile Regression. Econometric Society Monographs 38. Cambridge Univ. Press, Cambridge.
  • [17] Koenker, R. (2007). Quantile regression in R: A vignette. Available at http://cran.r-project.org.
  • [18] Koenker, R. and Bassett, G. J. (1978). Regression quantiles. Econometrica 46 33–50.
  • [19] Koltchinskii, V. (1997). M-estimation, convexity and quantiles. Ann. Statist. 25 435–477.
  • [20] Kong, L. and Mizera, I. (2008). Quantile tomography: Using quantiles with multivariate data. Preprint.
  • [21] Laine, B. (2001). Depth contours as multivariate quantiles: A directional approach. Unpublished Master dissertation (advisor M. Hallin). Univ. Libre de Bruxelles, Brussels.
  • [22] Liu, R. Y., Parelius, J. M. and Singh, K. (1999). Multivariate analysis by data depth: Descriptive statistics, graphics and inference. Ann. Statist. 27 783–840.
  • [23] Miller, K., Ramaswami, S., Rousseeuw, P., Sellarès, T., Souvaine, D., Streinu, I. and Struyf, A. (2003). Efficient computation of location depth contours by methods of computational geometry. Stat. Comput. 13 153–162.
  • [24] Mizera, I. (2002). On depth and deep points: A calculus. Ann. Statist. 30 1681–1736.
  • [25] Niemiro, W. (1992). Asymptotics for M-estimators defined by convex minimization. Ann. Statist. 20 1514–1533.
  • [26] Paindaveine, D. and Šiman, M. (2009). On directional multiple-output quantile regression. ECARES Working Paper 2009-011.
  • [27] Paindaveine, D. and Šiman, M. (2010). Computing multiple-output regression quantile regions. Submitted.
  • [28] Rousseeuw, P. J. and Ruts, I. (1996). Algorithm AS 307: Bivariate location depth. J. Appl. Stat. 45 516–526.
  • [29] Rousseeuw, P. J. and Hubert, M. (1999). Regression depth. J. Amer. Statist. Assoc. 94 388–433. With discussion and a reply by the authors and Stefan Van Aelst.
  • [30] Rousseeuw, P. J. and Ruts, I. (1999). The depth function of a population distribution. Metrika 49 213–244.
  • [31] Rousseeuw, P. J. and Struyf, A. (1998). Computing location depth and regression depth in higher dimensions. Stat. Comput. 8 193–203.
  • [32] Rousseeuw, P. J. and Struyf, A. (2004). Characterizing angular symmetry and regression symmetry. J. Statist. Plann. Inference 122 161–173. Contemporary data analysis: Theory and methods.
  • [33] Serfling, R. (2002). Quantile functions for multivariate analysis: Approaches and applications. Statist. Neerlandica 56 214–232. Special issue: Frontier research in theoretical statistics, 2000 (Eindhoven).
  • [34] Serfling, R. (2010). Equivariance and invariance properties of multivariate quantile and related functions, and the role of standardization. J. Nonparametr. Statist., in press.
  • [35] Struyf, A. and Rousseeuw, P. J. (2005). Halfspace depth and regression depth characterize the empirical distribution. J. Multivariate Anal. 69 135–153.
  • [36] Tukey, J. W. (1975). Mathematics and the picturing of data. In Proceedings of the International Congress of Mathematicians (Vancouver, B.C., 1974) 2 523–531. Canad. Math. Congress, Montreal.
  • [37] Wei, Y. (2008). An approach to multivariate covariate-dependent quantile contours with application to bivariate conditional growth charts. J. Amer. Statist. Assoc. 103 397–409.
  • [38] Wei, Y., Pere, A., Koenker, R. and He, X. (2005). Quantile regression methods for reference growth charts. Stat. Med. 25 1369–1382.
  • [39] Zuo, Y. and Serfling, R. (2000). General notions of statistical depth function. Ann. Statist. 28 461–482.
  • [40] Zuo, Y. and Serfling, R. (2000). Structural properties and convergence results for contours of sample statistical depth functions. Ann. Statist. 28 483–499.