Directional quantile regression in R

Boček, Pavel ; Šiman, Miroslav

Kybernetika, Tome 53 (2017) no. 3, pp. 480-492

Abstract (VO)

Recently, the eminently popular standard quantile regression has been generalized to the multiple-output regression setup by means of directional regression quantiles in two rather interrelated ways. Unfortunately, they lead to complicated optimization problems involving parametric programming, and this may be the main obstacle standing in the way of their wide dissemination. The presented R package modQR is intended to address this issue. It originates as a quite faithful translation of the authors' moQuantile toolbox for Octave and MATLAB, and provides all the necessary computational support for both the directional multiple-output quantile regression methods to the wide statistical public. The article offers a concise summary of the statistical theory behind modQR, overviews the package in brief, points out its departures from moQuantile, comments on its use and performance, and demonstrates its application.

Détail

DOI : 10.14736/kyb-2017-3-0480

Classification : 62-04, 62H05, 62J99, 65C60
Keywords: multivariate quantile; regression quantile; halfspace depth; regression depth; depth contour

Bibliographie

[1] Boček, P., Šiman, M.: modQR: Multiple-Output Directional Quantile Regression. R package version 0.1.0, 2015.

[2] Boček, P., Šiman, M.: Directional quantile regression in Octave and MATLAB. Kybernetika 52 (2016), 28-51. | DOI | MR

[3] Chakraborty, B.: On multivariate quantile regression. J. Statist. Planning Inference 110 (2003), 109-132. | DOI | MR

[4] Charlier, I., Paindaveine, D., Saracco, J.: Multiple-output regression through optimal quantization. ECARES Working Paper 2016-18.

[5] Chaudhury, P.: On a geometric notion of quantiles for multivariate data. J. Amer. Stat. Assoc. 91 (1996), 862-872. | DOI | MR

[6] Cheng, Y., Gooijer, J. G. De: On the $u$th geometric conditional quantile. J. Statist. Planning Inference 137 (2007), 1914-1930. | DOI | MR | Zbl

[7] Došlá, Š.: Conditions for bimodality and multimodality of a mixture of two unimodal densities. Kybernetika 45 (2009) 279-292. | MR | Zbl

[8] Hallin, M., Lu, Z., Paindaveine, D., Šiman, M.: Local bilinear multiple-output quantile/depth regression. Bernoulli 21 (2015), 1435-1466. | DOI | MR

[9] Hallin, M., Paindaveine, D., Šiman, M.: Multivariate quantiles and multiple-output regression quantiles: From ${L}_1$ optimization to halfspace depth. Ann. Statist. 38 (2010), 635-669. | DOI | MR

[10] Hallin, M., Paindaveine, D., Šiman, M.: Rejoinder. Ann. Statist. 38 (2010), 694-703. | DOI | MR

[11] Koenker, R.: Quantile Regression. Cambridge University Press, New York 2005. | DOI | MR | Zbl

[12] Koenker, R., Bassett, G. J.: Regression quantiles. Econometrica 46 (1978), 33-50. | DOI | MR | Zbl

[13] Koltchinskii, V.: ${M}$-estimation, convexity and quantiles. Ann. Statist. 25 (1997), 435-477. | DOI | MR

[14] Kong, L., Mizera, I.: Quantile tomography: Using quantiles with multivariate data. Statistica Sinica 22 (2012), 1589-1610. | DOI | MR

[15] McKeague, I. W., López-Pintado, S., Hallin, M., Šiman, M.: Analyzing growth trajectories. J. Developmental Origins of Health and Disease 2 (2011), 322-329. | DOI

[16] Paindaveine, D., Šiman, M.: On directional multiple-output quantile regression. J. Multivariate Anal. 102 (2011), 193-212. | DOI | MR | Zbl

[17] Paindaveine, D., Šiman, M.: Computing multiple-output regression quantile regions. Comput. Statist. Data Anal. 56 (2012), 840-853. | DOI | MR | Zbl

[18] Paindaveine, D., Šiman, M.: Computing multiple-output regression quantile regions from projection quantiles. Comput. Statist. 27 (2012), 29-49. | DOI | MR | Zbl

[19] Šiman, M.: On exact computation of some statistics based on projection pursuit in a general regression context. Commun. Statist. - Simulation and Computation 40 (2011), 948-956. | DOI | MR | Zbl

[20] Šiman, M.: Precision index in the multivariate context. Commun. Statist. - Theory and Methods 43 (2014), 377-387. | DOI | MR