Abstract:W e propose a ne w and general approach, the penali z ed asymptotic likelihood (Pal) maximi z ation, to joint parameter estimation and varia b le selection for models w here a likelihood or a convenient loss function is not readily availa b le. T he main idea is to fi rst construct a" likelihood" function b ased on an existing estimator and its asymptotic distrib ution function, and then maximiz e the constructed likelihood su b ject to some shrinkage penalty for a sparse estimation. I n this paper, w e focus on the general class of linear transformation models for survival data analysis, w hich include the Cox's proportional odds models and proportional ha z ards models as special cases. W e present ho w to construct a penali z ed asymptotic likelihood for linear transformation models, b ased on the estimation e q uation procedure developed b y Chen et al.(2002). T he ne w procedure leads to an estimator w ith improved e ffi-ciency and sparse representation. T heoretical properties of the Pal estimators are esta b lished, including the root-n consistency, selection consistency, and asymptotic normality. Furthermore, the ne w procedure enjoys easy implementation and its entire solution path can b eo b tained w ith a modifi ed LA RS algorithm. T he performance of the Pal estimator is illustrated b y simulated examples and the analysis of V eteran's A dministration lung cancer data.

Poisson Autoregressive Process Modeling Via the Penalized Conditional Maximum Likelihood Procedure

Simultaneous sparse model selection and coefficient estimation for heavy-tailed autoregressive processes

On the Maximum Penalized Full Likelihood Approach for Cox Model with Extreme Value for Heavily Censored Survival Data

Penalized Empirical Likelihood for Generalized Linear Models with Longitudinal Data.

Variable selection via penalized quasi-maximum likelihood method for spatial autoregressive model with missing response

Empirical likelihood-based inference in Poisson autoregressive model with conditional moment restrictions

PENALIZED VARIABLE SELECTION PROCEDURE FOR COX MODELS WITH SEMIPARAMETRIC RELATIVE RISK

Asymptotic Properties of Estimation in Generalized Partial Linear Mixed Models

Penalized empirical likelihood for high-dimensional generalized linear models with longitudinal data

Compound Poisson Processes, Latent Shrinkage Priors and Bayesian Nonconvex Penalization

On the Penalized Maximum Likelihood Estimation of High-Dimensional Approximate Factor Model

Penalized asymptotic likelihood approach for linear transformation model selection

Penalized Multiply Robust Estimation in High-Order Autoregressive Processes with Missing Explanatory Variables

Conditional L1 Estimation for Random Coefficient Integer-Valued Autoregressive Processes

Penalized Empirical Likelihood for Partially Linear Errors-in-variables Models

Estimation and Testing for a Poisson Autoregressive Model

Penalised inference for autoregressive moving average models with time-dependent predictors

Modeling Overdispersed or Underdispersed Count Data with Generalized Poisson Integer-Valued Autoregressive Processes

A penalized simulated maximum likelihood approach in parameter estimation for stochastic differential equations.

Adaptive lasso variable selection method for semiparametric spatial autoregressive panel data model with random effects

Maximum Likelihood Parameter Estimation for ARMAX Models Based on Stochastic Gradient Algorithm