Showing papers by "Runze Li published in 2007"

Tuning parameter selectors for the smoothly clipped absolute deviation method.

Abstract: SUMMARY The penalized least squares approach with smoothly clipped absolute deviation penalty has been consistently demonstrated to be an attractive regression shrinkage and selection method. It not only automatically and consistently selects the important variables, but also produces estimators which are as efficient as the oracle estimator. However, these attractive features depend on appropriate choice of the tuning parameter. We show that the commonly used generalized crossvalidation cannot select the tuning parameter satisfactorily, with a nonignorable overfitting effect in the resulting model. In addition, we propose a BIC tuning parameter selector, which is shown to be able to identify the true model consistently. Simulation studies are presented to support theoretical findings, and an empirical example is given to illustrate its use in the Female Labor Supply data.

Analysis of Longitudinal Data with Semiparametric Estimation of Covariance Function.

Abstract: Improving efficiency for regression coefficients and predicting trajectories of individuals are two important aspects in the analysis of longitudinal data. Both involve estimation of the covariance function. Yet challenges arise in estimating the covariance function of longitudinal data collected at irregular time points. A class of semiparametric models for the covariance function by that imposes a parametric correlation structure while allowing a nonparametric variance function is proposed. A kernel estimator for estimating the nonparametric variance function is developed. Two methods for estimating parameters in the correlation structure—a quasi-likelihood approach and a minimum generalized variance method—are proposed. A semiparametric varying coefficient partially linear model for longitudinal data is introduced, and an estimation procedure for model coefficients using a profile weighted least squares approach is proposed. Sampling properties of the proposed estimation procedures are studied, and asy...

A new estimation procedure for a partially nonlinear model via a mixed-effects approach.

Abstract: The authors consider the estimation of the parametric component of a partially nonlinear semiparametric regression model whose nonparametric component is viewed as a nuisance parameter. They show how estimation can proceed through a nonlinear mixed-effects model approach. They prove that under certain regularity conditions, the proposed estimate is consistent and asymptotically Gaussian. They investigate its finite-sample properties through simulations and illustrate its use with data on the relation between the photosynthetically active radiation and the net ecosystem-atmosphere exchange of carbon dioxide. Une nouvelle procedure fondee sur une approche a effets mixtes pour l'estimation des parametres d'un modele partiellement non lineaire Les auteurs s'interessent a l'estimation de la partie parametrique d'un modele de regression semiparametrique partiellement non lineaire dont la composante non parametrique est consideree comme nuisible. Ils montrent comment l'estimation est possible au moyen d'un modele non lineaire a effets mixtes. Ils demontrent que sous certaines conditions de regularite, l'estimateur propose est convergent et asymptotiquement gaussien. Ils en etudient le comportement a taille finie au moyen de simulations et en illustrent l'emploi a l'aide de donnees concernant le rayonnement absorbe par photosynthese en relation avec le bilan des echanges en bioxyde de carbone entre l'atmosphere et l'ecosysteme.

Empirical Kriging models and their applications to QSAR

Abstract: A general Kriging model consists of two additive components: a parametric term and a stochastic error process. It is known that Kriging is an interpolating predictor and allows for a better fit to the data, but suffers from a decreasing ability to generalize to unseen data. By incorporating a disturbing or an independent random error term into Kriging model, the resulting model, which is called empirical Kriging model in the literature, may provide more accurate prediction for the highly noisy data than the Kriging model. This paper presents an extensive survey of the empirical Kriging model for quantitative structure-activity relationship (QSAR) research and extends the parameters estimation technique with highly efficiency. In addiction, QSAR models are established by combining Kriging model or empirical Kriging model with principal components regression (PCR) and partial least squares regression (PLSR). We demonstrate for the real data set that the suggested empirical Kriging model can significantly improve the prediction ability of some commonly used models, including the Kriging model. Copyright © 2007 John Wiley & Sons, Ltd.