On the solution of the errors in variables problem using thel 1 norm

A fundamental problem in data analysis is that of fitting a given model to observed data. It is commonly assumed that only the dependent variable values are in error, and the least squares criterion is often used to fit the model. When significant errors occur in all the variables, then an alternative approach which is frequently suggested for this errors in variables problem is to minimize the sum of squared orthogonal distances between each data point and the curve described by the model equation. It has long been recognized that the use of least squares is not always satisfactory, and thel ₁ criterion is often superior when estimating the true form of data which contain some very inaccurate observations. In this paper the measure of goodness of fit is taken to be thel ₁ norm of the errors. A Levenberg-Marquardt method is proposed, and the main objective is to take full advantage of the structure of the subproblems so that they can be solved efficiently.     

On same-realization prediction in an infinite-order autoregressive process

Let observations come from an infinite-order autoregressive (AR) process. For predicting the future of the observed time series (referred to as the same-realization prediction), we use the least-squares predictor obtained by fitting a finite-order AR model. We also allow the order to become infinite as the number of observations does in order to obtain a better approximation. Moment bounds for the inverse sample covariance matrix with an increasing dimension are established under various conditions. We then apply these results to obtain an asymptotic expression for the mean-squared prediction error of the least-squares predictor in same-realization and increasing-order settings. The second-order term of this expression is the sum of two terms which measure both the goodness of fit and model complexity. It forms the foundation for a companion paper by Ing and Wei (Order selection for same-realization predictions in autoregressive processes, Technical report C-00-09, Institute of Statistical Science, Academia Sinica, Taipei, Taiwan, ROC, 2000) which provides the first theoretical verification that AIC is asymptotically efficient for same-realization predictions. Finally, some comparisons between the least-squares predictor and the ridge regression predictor are also given.     

Conditional and unconditional methods for selecting variables in linear mixed models

In the problem of selecting the explanatory variables in the linear mixed model, we address the derivation of the (unconditional or marginal) Akaike information criterion (AIC) and the conditional AIC (cAIC). The covariance matrices of the random effects and the error terms include unknown parameters like variance components, and the selection procedures proposed in the literature are limited to the cases where the parameters are known or partly unknown. In this paper, AIC and cAIC are extended to the situation where the parameters are completely unknown and they are estimated by the general consistent estimators including the maximum likelihood (ML), the restricted maximum likelihood (REML) and other unbiased estimators. We derive, related to AIC and cAIC, the marginal and the conditional prediction error criteria which select superior models in light of minimizing the prediction errors relative to quadratic loss functions. Finally, numerical performances of the proposed selection procedures are investigated through simulation studies.     

Optimal prediction for linear regression with infinitely many parameters

The problem of optimal prediction in the stochastic linear regression model with infinitely many parameters is considered. We suggest a prediction method that outperforms asymptotically the ordinary least squares predictor. Moreover, if the random errors are Gaussian, the method is asymptotically minimax over ellipsoids in ?₂. The method is based on a regularized least squares estimator with weights of the Pinsker filter. We also consider the case of dynamic linear regression, which is important in the context of transfer function modeling.     

Nonparametric lack-of-fit tests for parametric mean-regression models with censored data

We developed two kernel smoothing based tests of a parametric mean-regression model against a nonparametric alternative when the response variable is right-censored. The new test statistics are inspired by the synthetic data and the weighted least squares approaches for estimating the parameters of a (non)linear regression model under censoring. The asymptotic critical values of our tests are given by the quantiles of the standard normal law. The tests are consistent against fixed alternatives, local Pitman alternatives and uniformly over alternatives in Hölder classes of functions of known regularity.     

A two-stage approach to semilinear in-slide models

The semilinear in-slide models (SLIMs) have been shown to be effective methods for normalizing microarray data [J. Fan, P. Tam, G. Vande Woude, Y. Ren, Normalization and analysis of cDNA micro-arrays using within-array replications applied to neuroblastoma cell response to a cytokine, Proceedings of the National Academy of Science (2004) 1135-1140]. Using a backfitting method, [J. Fan, H. Peng, T. Huang, Semilinear high-dimensional model for normalization of microarray data: a theoretical analysis and partial consistency, Journal of American Statistical Association, 471, (2005) 781-798] proposed a profile least squares (PLS) estimation for the parametric and nonparametric components. The general asymptotic properties for their estimator is not developed. In this paper, we consider a new approach, two-stage estimation, which enables us to establish the asymptotic normalities for both of the parametric and nonparametric component estimators. We further propose a plug-in bandwidth selector using the asymptotic normality of the nonparametric component estimator. The proposed method allow for the modeling of the aggregated SLIMs case where we can explicitly show that taking the aggregated information into account can improve both of the parametric and nonparametric component estimator by the proposed two-stage approach. Some simulation studies are conducted to illustrate the finite sample performance of the proposed procedures.     

Offline and online weighted least squares estimation of nonstationary power ARCH processes

This paper proposes two estimation methods based on a weighted least squares criterion for non-(strictly) stationary power ARCH models. The weights are the squared volatilities evaluated at a known value in the parameter space. The first method is adapted for fixed sample size data while the second one allows for online data available in real time. It will be shown that these methods provide consistent and asymptotically Gaussian estimates having asymptotic variance equal to that of the quasi-maximum likelihood estimate (QMLE) regardless of the value of the weighting parameter. Finite-sample performances of the proposed WLS estimates are shown via a simulation study for various sub-classes of power ARCH models.     

Robust Bayesian prediction and estimation under a squared log error loss function

Robust Bayesian analysis is concerned with the problem of making decisions about some future observation or an unknown parameter, when the prior distribution belongs to a class Γ instead of being specified exactly. In this paper, the problem of robust Bayesian prediction and estimation under a squared log error loss function is considered. We find the posterior regret Γ-minimax predictor and estimator in a general class of distributions. Furthermore, we construct the conditional Γ-minimax, most stable and least sensitive prediction and estimation in a gamma model. A prequential analysis is carried out by using a simulation study to compare these predictors.     

An overview on the shrinkage properties of partial least squares regression

The aim of this paper is twofold. In the first part, we recapitulate the main results regarding the shrinkage properties of partial least squares (PLS) regression. In particular, we give an alternative proof of the shape of the PLS shrinkage factors. It is well known that some of the factors are >1. We discuss in detail the effect of shrinkage factors for the mean squared error of linear estimators and argue that we cannot extend the results to PLS directly, as it is nonlinear. In the second part, we investigate the effect of shrinkage factors empirically. In particular, we point out that experiments on simulated and real world data show that bounding the absolute value of the PLS shrinkage factors by 1 seems to leads to a lower mean squared error.     

Sparse estimation in functional linear regression

As a useful tool in functional data analysis, the functional linear regression model has become increasingly common and been studied extensively in recent years. In this paper, we consider a sparse functional linear regression model which is generated by a finite number of basis functions in an expansion of the coefficient function. In this model, we do not specify how many and which basis functions enter the model, thus it is not like a typical parametric model where predictor variables are pre-specified. We study a general framework that gives various procedures which are successful in identifying the basis functions that enter the model, and also estimating the resulting regression coefficients in one-step. We adopt the idea of variable selection in the linear regression setting where one adds a weighted L₁ penalty to the traditional least squares criterion. We show that the procedures in our general framework are consistent in the sense of selecting the model correctly, and that they enjoy the oracle property, meaning that the resulting estimators of the coefficient function have asymptotically the same properties as the oracle estimator which uses knowledge of the underlying model. We investigate and compare several methods within our general framework, via a simulation study. Also, we apply the methods to the Canadian weather data.     

Generalized Levinson-Durbin sequences, binomial coefficients and autoregressive estimation

For a discrete time second-order stationary process, the Levinson-Durbin recursion is used to determine the coefficients of the best linear predictor of the observation at time k+1, given k previous observations, best in the sense of minimizing the mean square error. The coefficients determined by the recursion define a Levinson-Durbin sequence. We also define a generalized Levinson-Durbin sequence and note that binomial coefficients form a special case of a generalized Levinson-Durbin sequence. All generalized Levinson-Durbin sequences are shown to obey summation formulas which generalize formulas satisfied by binomial coefficients. Levinson-Durbin sequences arise in the construction of several autoregressive model coefficient estimators. The least squares autoregressive estimator does not give rise to a Levinson-Durbin sequence, but least squares fixed point processes, which yield least squares estimates of the coefficients unbiased to order 1/T, where T is the sample length, can be combined to construct a Levinson-Durbin sequence. By contrast, analogous fixed point processes arising from the Yule-Walker estimator do not combine to construct a Levinson-Durbin sequence, although the Yule-Walker estimator itself does determine a Levinson-Durbin sequence. The least squares and Yule-Walker fixed point processes are further studied when the mean of the process is a polynomial time trend that is estimated by least squares.     

A family of estimators for multivariate kurtosis in a nonnormal linear regression model

In this paper, we propose a new estimator for a kurtosis in a multivariate nonnormal linear regression model. Usually, an estimator is constructed from an arithmetic mean of the second power of the squared sample Mahalanobis distances between observations and their estimated values. The estimator gives an underestimation and has a large bias, even if the sample size is not small. We replace this squared distance with a transformed squared norm of the Studentized residual using a monotonic increasing function. Our proposed estimator is defined by an arithmetic mean of the second power of these squared transformed squared norms with a correction term and a tuning parameter. The correction term adjusts our estimator to an unbiased estimator under normality, and the tuning parameter controls the sizes of the squared norms of the residuals. The family of our estimators includes estimators based on ordinary least squares and predicted residuals. We verify that the bias of our new estimator is smaller than usual by constructing numerical experiments.     

Multiple criteria linear regression

The unknown parameters in multiple linear regression models may be estimated using any one of a number of criteria such as the minimization of the sum of squared errors MSSE, the minimization of the sum of absolute errors MSAE, and the minimization of the maximum absolute error MMAE. At present, the MSSE or the least squares criterion continues to be the most popular. However, at times the choice of a criterion is not clear from statistical, practical or other considerations. Under such circumstances, it may be more appropriate to use multiple criteria rather than a single criterion to estimate the unknown parameters in a multiple linear regression model. We motivate the use of multiple criteria estimation in linear regression models with an example, propose a few models, and outline a solution procedure.     

James-Stein estimators for time series regression models

The least squares (LS) estimator seems the natural estimator of the coefficients of a Gaussian linear regression model. However, if the dimension of the vector of coefficients is greater than 2 and the residuals are independent and identically distributed, this conventional estimator is not admissible. James and Stein [Estimation with quadratic loss, Proceedings of the Fourth Berkely Symposium vol. 1, 1961, pp. 361-379] proposed a shrinkage estimator (James-Stein estimator) which improves the least squares estimator with respect to the mean squares error loss function. In this paper, we investigate the mean squares error of the James-Stein (JS) estimator for the regression coefficients when the residuals are generated from a Gaussian stationary process. Then, sufficient conditions for the JS to improve the LS are given. It is important to know the influence of the dependence on the JS. Also numerical studies illuminate some interesting features of the improvement. The results have potential applications to economics, engineering, and natural sciences.     

Autoregressive frequency detection using Regularized Least Squares

Tracking of an unknown frequency embedded in noise is widely applied in a variety of applications. Unknown frequencies can be obtained by approximating generalized spectral density of a periodic process by an autoregressive (AR) model. The advantage is that an AR model has a simple structure and its parameters can be easily estimated iteratively, which is crucial for online (real-time) applications. Typically, the order of the AR approximation is chosen by information criteria. However, with an increase of a sample size, model order may change, which leads to re-estimation of all model parameters. We propose a new iterative procedure for frequency detection based on a regularization of an empirical information matrix. The suggested method enables to avoid the repeated model selection as well as parameter estimation steps and therefore optimize computational costs. The asymptotic properties of the proposed regularized AR (RAR) frequency estimates are derived and performance of RAR is evaluated by numerical examples.     

An adaptive estimation of MAVE

Minimum average variance estimation (MAVE, Xia et al. (2002) [29]) is an effective dimension reduction method. It requires no strong probabilistic assumptions on the predictors, and can consistently estimate the central mean subspace. It is applicable to a wide range of models, including time series. However, the least squares criterion used in MAVE will lose its efficiency when the error is not normally distributed. In this article, we propose an adaptive MAVE which can be adaptive to different error distributions. We show that the proposed estimate has the same convergence rate as the original MAVE. An EM algorithm is proposed to implement the new adaptive MAVE. Using both simulation studies and a real data analysis, we demonstrate the superior finite sample performance of the proposed approach over the existing least squares based MAVE when the error distribution is non-normal and the comparable performance when the error is normal.     

On Least Squares Estimation for Stable Nonlinear AR Processes

Following a Markov chain approach, this paper establishes asymptotic properties of the least squares estimator in nonlinear autoregressive (NAR) models. Based on conditions ensuring the stability of the model and allowing the use of a strong law of large number for a wide class of functions, our approach improves some known results on strong consistency and asymptotic normality of the estimator. The exact convergence rate is established by a law of the iterated logarithm. Based on this law and a generalized Akaike's information criterion, we build a strongly consistent procedure for selection of NAR models. Detailed results are given for familiar nonlinear AR models like exponential AR models, threshold models or multilayer feedforward perceptions.     

On prediction rate in partial functional linear regression

We consider a prediction of a scalar variable based on both a function-valued variable and a finite number of real-valued variables. For the estimation of the regression parameters, which include the infinite dimensional function as well as the slope parameters for the real-valued variables, it is inevitable to impose some kind of regularization. We consider two different approaches, which are shown to achieve the same convergence rate of the mean squared prediction error under respective assumptions. One is based on functional principal components regression (FPCR) and the alternative is functional ridge regression (FRR) based on Tikhonov regularization. Also, numerical studies are carried out for a simulation data and a real data.     

Second-order nonlinear least squares estimation

The ordinary least squares estimation is based on minimization of the squared distance of the response variable to its conditional mean given the predictor variable. We extend this method by including in the criterion function the distance of the squared response variable to its second conditional moment. It is shown that this “second-order” least squares estimator is asymptotically more efficient than the ordinary least squares estimator if the third moment of the random error is nonzero, and both estimators have the same asymptotic covariance matrix if the error distribution is symmetric. Simulation studies show that the variance reduction of the new estimator can be as high as 50% for sample sizes lower than 100. As a by-product, the joint asymptotic covariance matrix of the ordinary least squares estimators for the regression parameter and for the random error variance is also derived, which is only available in the literature for very special cases, e.g. that random error has a normal distribution. The results apply to both linear and nonlinear regression models, where the random error distributions are not necessarily known.