Nonparametric Variable Transformation in Sufficient Dimension Reduction

Sufficient dimension reduction (SDR) techniques have proven to be very useful data analysis tools in various applications. Underlying many SDR techniques is a critical assumption that the predictors are elliptically contoured. When this assumption appears to be wrong, practitioners usually try variable transformation such that the transformed predictors become (nearly) normal. The transformation function is often chosen from the log and power transformation family, as suggested in the celebrated Box–Cox model. However, any parametric transformation can be too restrictive, causing the danger of model misspecification. We suggest a nonparametric variable transformation method after which the predictors become normal. To demonstrate the main idea, we combine this flexible transformation method with two well-established SDR techniques, sliced inverse regression (SIR) and inverse regression estimator (IRE). The resulting SDR techniques are referred to as TSIR and TIRE, respectively. Both simulation and real data results show that TSIR and TIRE have very competitive performance. Asymptotic theory is established to support the proposed method. The technical proofs are available as supplementary materials.

[1]  Larry A. Wasserman,et al.  The Nonparanormal: Semiparametric Estimation of High Dimensional Undirected Graphs , 2009, J. Mach. Learn. Res..

[2]  Michael I. Jordan,et al.  Kernel dimension reduction in regression , 2009, 0908.1854.

[3]  Lixing Zhu,et al.  On Sliced Inverse Regression With High-Dimensional Covariates , 2006 .

[4]  W. Fung,et al.  DIMENSION REDUCTION BASED ON CANONICAL CORRELATION , 2002 .

[5]  Lixing Zhu,et al.  Sufficient dimension reduction through discretization-expectation estimation , 2010 .

[6]  Yingcun Xia,et al.  Sliced Regression for Dimension Reduction , 2008 .

[7]  Hui Zou,et al.  Semiparametric Sparse Discriminant Analysis , 2013 .

[8]  B. Li,et al.  Dimension reduction for nonelliptically distributed predictors , 2009, 0904.3842.

[9]  J. Brian Gray,et al.  Applied Regression Including Computing and Graphics , 1999, Technometrics.

[10]  Liping Zhu,et al.  A Semiparametric Approach to Dimension Reduction , 2012, Journal of the American Statistical Association.

[11]  H. Tong,et al.  Article: 2 , 2002, European Financial Services Law.

[12]  R. Cook,et al.  Principal Hessian Directions Revisited , 1998 .

[13]  R. Cook,et al.  Reweighting to Achieve Elliptically Contoured Covariates in Regression , 1994 .

[14]  S. Weisberg,et al.  Comments on "Sliced inverse regression for dimension reduction" by K. C. Li , 1991 .

[15]  R. Cook,et al.  Optimal sufficient dimension reduction in regressions with categorical predictors , 2002 .

[16]  R. Cook,et al.  Likelihood-Based Sufficient Dimension Reduction , 2009 .

[17]  R. Dennis Cook,et al.  Estimating central subspaces via inverse third moments , 2003 .

[18]  Bing Li,et al.  Dimension reduction for non-elliptically distributed predictors: second-order methods , 2010 .

[19]  Shaoli Wang,et al.  On Directional Regression for Dimension Reduction , 2007 .

[20]  Yuexiao Dong.  Dimension reduction for non-elliptically distributed predictors , 2009 .

[21]  R. Cook,et al.  Dimension reduction for conditional mean in regression , 2002 .

[22]  Efstathia Bura,et al.  Moment-based dimension reduction for multivariate response regression , 2006 .

[23]  Ker-Chau Li,et al.  On Principal Hessian Directions for Data Visualization and Dimension Reduction: Another Application of Stein's Lemma , 1992 .

[24]  Rolph E. Anderson,et al.  Canonical Correlation , 2016 .

[25]  Ker-Chau Li,et al.  Sliced Inverse Regression for Dimension Reduction , 1991 .

[26]  Chih-Ling Tsai,et al.  Contour Projected Dimension Reduction , 2008, 0911.3481.

[27]  Xuming He,et al.  Dimension reduction based on constrained canonical correlation and variable filtering , 2008, 0808.0977.

[28]  Yi Lin,et al.  Discriminant analysis through a semiparametric model , 2003 .

[29]  Sylvain Girard.  Sliced Inverse Regression , 2014 .

[30]  R. Dennis Cook,et al.  A Model-Free Test for Reduced Rank in Multivariate Regression , 2003 .

[31]  R. Dennis Cook,et al.  Testing predictor contributions in sufficient dimension reduction , 2004, math/0406520.

[32]  C. Klaassen,et al.  Efficient estimation in the bivariate normal copula model: normal margins are least favourable , 1997 .

[33]  R. Cook,et al.  Sufficient Dimension Reduction via Inverse Regression , 2005 .