Abstract
We propose a robust estimation procedure based on local Walsh-average regression (LWR) for single-index models. Our novel method provides a root-n consistent estimate of the single-index parameter under some mild regularity conditions; the estimate of the unknown link function converges at the usual rate for the nonparametric estimation of a univariate covariate. We theoretically demonstrate that the new estimators show significant efficiency gain across a wide spectrum of non-normal error distributions and have almost no loss of efficiency for the normal error. Even in the worst case, the asymptotic relative efficiency (ARE) has a lower bound compared with the least squares (LS) estimates; the lower bounds of the AREs are 0.864 and 0.8896 for the single-index parameter and nonparametric function, respectively. Moreover, the ARE of the proposed LWR-based approach versus the ARE of the LS-based method has an expression that is closely related to the ARE of the signed-rank Wilcoxon test as compared with the t-test. In addition, to obtain a sparse estimate of the single-index parameter, we develop a variable selection procedure by combining the estimation method with smoothly clipped absolute deviation penalty; this procedure is shown to possess the oracle property. We also propose a Bayes information criterion (BIC)-type criterion for selecting the tuning parameter and further prove its ability to consistently identify the true model. We conduct some Monte Carlo simulations and a real data analysis to illustrate the finite sample performance of the proposed methods.
Similar content being viewed by others
References
Carroll R J, Fan J, Gijbels I, et al. Generalized partially linear single-index models. J Amer Statist Assoc, 1997, 92: 477–489
Fan J, Li R. Variable selection via nonconcave penalized likelihood and its oracle properties. J Amer Statist Assoc, 2001, 96: 1348–1360
Fan Y, Härdle W, Wang W, et al. Composite quantile regression for the single-index model. SFB 649 Discussion paper 2013–010. Sonderforschungsbereich 649. Berlin: Humboldt University, 2013, https://doi.org/10.2139/ssrn.2892647
Feng L, Zou C, Wang Z. Local Walsh-average regression. J Multivariate Anal, 2012, 106: 36–48
Feng L, Zou C, Wang Z. Rank-based inference for the single-index model. Statist Probab Lett, 2012, 82: 535–541
Feng S, Xue L. Model detection and estimation for single-index varying coefficient model. J Multivariate Anal, 2015, 139: 227–244
Härdle W, Hall P, Ichimura H. Optimal smoothing in single-index models. Ann Statist, 1993, 21: 157–178
Härdle W, Stoker T M. Investigating smooth multiple regression by method of average derivatives. J Amer Statist Assoc, 1989, 84: 986–995
Hettmansperger T P, McKean J W. Robust Nonparametric Statistical Methods, 2nd ed. New York: Chapman-Hall, 2011
Hodges J L, Lehmann E L. The efficiency of some nonparametric competitors of the t-test. Ann Math Statist, 1956, 27: 324–335
Hristache M, Juditski A, Spokoiny V. Direct estimation of the index coefficients in a single-index model. Ann Statist, 2001, 29: 595–623
Kai B, Li R, Zou H. Local composite quantile regression smoothing: An efficient and safe alternative to local polynomial regression. J R Stat Soc Ser B Stat Methodol, 2010, 72: 49–69
Kong E, Xia Y. Variable selection for the single-index model. Biometrika, 2007, 94: 217–229
Kong E, Xia Y. A single-index quantile regression model and its estimation. Econom Theory, 2012, 28: 730–768
Luo S, Ghosal S. Forward selection and estimation in high dimensional single index models. Stat Methodol, 2016, 33: 172–179
Neykov M, Liu S, Cai T. Li-regularized least squares for support recovery of high dimensional single index models with Gaussian designs. J Mach Learn Res, 2016, 17: 1–37
Peng H, Huang T. Penalized least squares for single index models. J Statist Plann Infer, 2011, 141: 1362–1379
Radchenko P. High dimensional single index models. J Multivariate Anal, 2015, 139: 266–282
Shang S, Zou C, Wang Z. Local Walsh-average regression for semiparametric varying-coefficient models. Statist Probab Lett, 2012, 82: 1815–1822
Terpstra J, McKean J. Rank-based analysis of linear models using R. J Statist Soft, 2005, 14: 1–26
Tian G, Wang M, Song L. Variable selection in the high-dimensional continuous generalized linear model with current status data. J Appl Stat, 2014, 41: 467–483
Wang G, Wang L. Spline estimation and variable selection for single-index prediction models with diverging number of index parameters. J Statist Plann Infer, 2015, 162: 1–19
Wang H. Bayesian estimation and variable selection for single index models. Comput Statist Data Anal, 2009, 53: 2617–2627
Wang H, Leng C. Unified lasso estimation via least squares approximation. J Amer Statist Assoc, 2007, 101: 1418–1429
Wang K. Robust direction identification and variable selection in high dimensional general single-index models. J Korean Statist Soc, 2015, 44: 606–618
Wang L. Wilcoxon-type generalized Bayesian information criterion. Biometrika, 2009, 96: 163–173
Wang L, Kai B, Li R. Local rank inference for varying coefficient models. J Amer Statist Assoc, 2009, 488: 1631–1645
Wang M, Song L, Tian G. SCAD-penalized least absolute deviation regression in high dimensional models. Comm Statist Theory Meth, 2015, 44: 2452–2472
Wang M, Song L, Wang X. Quadratic approximation via the SCAD penalty with a diverging number of parameters. Comm Statist Simulation Comput, 2016, 45: 1–16
Wang M, Tian G. Robust group non-convex estimations for high-dimensional partially linear models. J Nonparametr Stat, 2016, 28: 49–67
Wang M, Wang X. Adaptive LASSO estimators for ultrahigh dimensional generalized linear models. Statist Probab Lett, 2014, 89: 41–50
Wang T, Xu P, Zhu L. Non-convex penalized estimation in high-dimensional models with single-index structure. J Multivariate Anal, 2012, 109: 221–235
Wang T, Zhu L. A distribution-based LASSO for a general single-index model. Sci China Math, 2015, 58: 109–130
Wang X, Zhao S, Wang M. Restricted profile estimation for partially linear models with large-dimensional covariates. Statist Probab Lett, 2017, 128: 71–76
Wu T Z, Yu K, Yu Y. Single-index quantile regression. J Multivariate Anal, 2010, 101: 1607–1621
Xia Y. A constructive approach to the estimation of dimension reduction directions. Ann Statist, 2007, 35: 2654–2690
Xia Y, Härdle W. Semi-parametric estimation of partially linear single-index models. J Multivariate Anal, 2006, 97: 1162–1184
Xia Y, Tong H, Li W K, et al. An adaptive estimation of dimension reduction space. J R Stat Soc Ser B Stat Methodol, 2002, 64: 363–410
Zeng P, He T, Zhu Y. An LASSO-type approach for estimation and variable selection in single index models. J Comput Graph Statist, 2012, 21: 92–109
Zhu L, Huang M, Li R. Semiparametric quantile regression with high-dimensional covariates. Statist Sinica, 2012, 22: 1379–1401
Zhu L, Qian L, Lin J. Variable selection in a class of single-index models. Ann Inst Statist Math, 2011, 63: 1277–1293
Zhu L, Zhu L. Nonconcave penalized inverse regression in single-index models with high dimensional predictors. J Multivariate Anal, 2009, 100: 862–875
Zou H, Li R. One-step sparse estimates in nonconcave penalized likelihood models. Ann Statist, 2008, 36: 1509–1533
Acknowledgements
This work was partially supported by National Natural Science Foundation of China (Grant Nos. 11801168, 11801169, 11571055 and 11671059) and the Natural Science Foundation of Hunan Province (Grant No. 2018JJ3322). The authors are grateful to the two anonymous referees whose comments led to a significant improvement of the paper.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Yang, J., Lu, F. & Yang, H. Local Walsh-average-based estimation and variable selection for single-index models. Sci. China Math. 62, 1977–1996 (2019). https://doi.org/10.1007/s11425-017-9262-3
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11425-017-9262-3
Keywords
- single-index models
- local Walsh-average regression
- asymptotic relative efficiency
- variable selection
- oracle property