Abstract:The additive partially linear model (APLM) combines the flexibility of nonparametric regression with the parsimony of regression models, and has been widely used as a popular tool in multivariate nonparametric regression to alleviate the "curse of dimensionality". A natural question raised in practice is the choice of structure in the nonparametric part, that is, whether the continuous covariates enter into the model in linear or nonparametric form. In this paper, we present a comprehensive framework for simultaneous sparse model identification and learning for ultra-high-dimensional APLMs where both the linear and nonparametric components are possibly larger than the sample size. We propose a fast and efficient two-stage procedure. In the first stage, we decompose the nonparametric functions into a linear part and a nonlinear part. The nonlinear functions are approximated by constant spline bases, and a triple penalization procedure is proposed to select nonzero components using adaptive group LASSO. In the second stage, we refit data with selected covariates using higher order polynomial splines, and apply spline-backfitted local-linear smoothing to obtain asymptotic normality for the estimators. The procedure is shown to be consistent for model structure identification. It can identify zero, linear, and nonlinear components correctly and efficiently. Inference can be made on both linear coefficients and nonparametric functions. We conduct simulation studies to evaluate the performance of the method and apply the proposed method to a dataset on the Shoot Apical Meristem (SAM) of maize genotypes for illustration.

Identification of partially linear structure in additive models with an application to gene expression prediction from sequences

Sparse Model Identification and Learning for Ultra-high-dimensional Additive Partially Linear Models

NONPARAMETRIC IDENTIFICATION AND ESTIMATION OF A GENERALIZED ADDITIVE MODEL WITH A FLEXIBLE ADDITIVE STRUCTURE AND UNKNOWN LINK

Partially Linear Additive Models with Unknown Link Functions

Efficient Estimation of Additive Partially Linear Models

Determination of linear components in additive models

A Robust Penalized Estimation for Identification in Semiparametric Additive Models

Discovering Model Structure for Partially Linear Models

A KERNEL-BASED METHOD FOR ESTIMATING ADDITIVE PARTIALLY LINEAR MODELS

ADDITIVE COEFFICIENT MODELING VIA POLYNOMIAL SPLINE

Identification for Semiparametric Varying Coefficient Partially Linear Models

More Efficient Estimation of Multivariate Additive Models Based on Tensor Decomposition and Penalization.

Robust and efficient estimator for simultaneous model structure identification and variable selection in generalized partial linear varying coefficient models with longitudinal data

Component Selection in the Additive Regression Model

Automatic Variable Selection for Partially Linear Functional Additive Model and Its Application to the Tecator Data Set

Partially Linear Additive Functional Regression

Scad-Penalised Generalised Additive Models with Non-Polynomial Dimensionality

A robust and efficient estimation and variable selection method for partially linear models with large-dimensional covariates

Estimation and model selection in generalized additive partial linear models for correlated data with diverging number of covariates

Statistical Inference on Semi-Parametric Partial Linear Additive Models

Decomposition and Identification of Linear Structural Equation Models