Infeasible Estimator Research Articles

Oracle-efficient estimation for the mean function of missing covariate data based on noparametrically estimated selection probabilities

A weighted local linear estimator for the mean function is constructed based on nonparametrically estimated selection probabilities when covariates are missing at random. The weighted estimator is shown to be oracle-efficient, meaning it is uniformly as efficient as the infeasible estimator that uses the true selection probabilities assumed to be known as a prior. By applying oracle efficiency and the extreme value distribution of the weighted local linear estimator, asymptotically accurate simultaneous confidence bands for the mean function are derived. This approach avoids the risk of model misspecification for the missingness mechanism. Simulation studies are conducted to examine the finite sample performance, which supports the asymptotic results. The proposed methods are illustrated through the analysis of two real data sets.

Model selection and model averaging for matrix exponential spatial models

In this paper, we focus on a model specification problem in spatial econometric models when an empiricist needs to choose from a pool of candidates for the spatial weights matrix. We propose a model selection (MS) procedure for the matrix exponential spatial specification (MESS), when the true spatial weights matrix may not be in the set of candidate spatial weights matrices. We show that the selection estimator is asymptotically optimal in the sense that asymptotically it is as efficient as the infeasible estimator that uses the best candidate spatial weights matrix. The proposed selection procedure is also consistent in the sense that when the data generating process involves spatial effects, it chooses the true spatial weights matrix with probability approaching one in large samples. We also propose a model averaging (MA) estimator that compromises across a set of candidate models. We show that it is asymptotically optimal. We further flesh out how to extend the proposed selection and averaging schemes to higher order specifications and to the MESS with heteroscedasticity. Our Monte Carlo simulation results indicate that the MS and MA estimators perform well in finite samples. We also illustrate the usefulness of the proposed MS and MA schemes in a spatially augmented economic growth model.

Simultaneous confidence bands for comparing variance functions of two samples based on deterministic designs

Asymptotically correct simultaneous confidence bands (SCBs) are proposed in both multiplicative and additive form to compare variance functions of two samples in the nonparametric regression model based on deterministic designs. The multiplicative SCB is based on two-step estimation of ratio of the variance functions, which is as efficient, up to order $$n^{-1/2}$$ , as an infeasible estimator if the two mean functions are known a priori. The additive SCB, which is the log transform of the multiplicative SCB, is location and scale invariant in the sense that the width of SCB is free of the unknown mean and variance functions of both samples. Simulation experiments provide strong evidence that corroborates the asymptotic theory. The proposed SCBs are used to analyze several strata pressure data sets from the Bullianta Coal Mine in Erdos City, Inner Mongolia, China.

Efficiency gains in least squares estimation: A new approach

In pursuit of efficiency, we propose a new way to construct least squares estimators, as the minimizers of an augmented objective function that takes explicitly into account the variability of the error term and the resulting uncertainty, as well as the possible existence of heteroskedasticity. We initially derive an infeasible estimator which we then approximate using Ordinary Least Squares (OLS) residuals from a first-step regression to obtain the feasible “HOLS” estimator. This estimator has negligible bias, is consistent and outperforms OLS in terms of finite-sample Mean Squared Error, but also in terms of asymptotic efficiency, under all skedastic scenarios, including homoskedasticity. Analogous efficiency gains are obtained for the case of Instrumental Variables estimation. Theoretical results are accompanied by simulations that support them.

Estimation of the Robin coefficient field in a Poisson problem with uncertain conductivity field**This work was partially supported by NSF grants ACI-1550547 and 1654311.

We consider the reconstruction of a heterogeneous coefficient field in a Robin boundary condition on an inaccessible part of the boundary in a Poisson problem with an uncertain (or unknown) inhomogeneous conductivity field in the interior of the domain. To account for model errors that stem from the uncertainty in the conductivity coefficient, we treat the unknown conductivity as a nuisance parameter and carry out approximative premarginalization over it, and invert for the Robin coefficient field only. We approximate the related modelling errors via the Bayesian approximation error (BAE) approach. The uncertainty analysis presented here relies on a local linearization of the parameter-to-observable map at the maximum a posteriori (MAP) estimates, which leads to a normal (Gaussian) approximation of the parameter posterior density. To compute the MAP point we apply an inexact Newton conjugate gradient approach based on the adjoint methodology. The construction of the covariance is made tractable by invoking a low-rank approximation of the data misfit component of the Hessian. Two numerical experiments are considered: one where the prior covariance on the conductivity is isotropic, and one where the prior covariance on the conductivity is anisotropic. Results are compared to those based on standard error models, with particular emphasis on the feasibility of the posterior uncertainty estimates. We show that the BAE approach is a feasible one in the sense that the predicted posterior uncertainty is consistent with the actual estimation errors, while neglecting the related modelling error yields infeasible estimates for the Robin coefficient. In addition, we demonstrate that the BAE approach is approximately as computationally expensive (measured in the number of PDE solves) as the conventional error approach.

Improved Calibration Method for Dynamic Traffic Assignment Models: Constrained Extended Kalman Filter

The calibration of dynamic traffic assignment (DTA) models involves the estimation of model parameters to best replicate real-world measurements. Good calibration is essential to estimate and predict accurately traffic states, which are crucial for traffic management applications to alleviate congestion. A widely used approach to calibrate simulation-based DTA models is the extended Kalman filter (EKF). The EKF assumes that the DTA model parameters are unconstrained, although they are in fact constrained; for instance, origin–destination (O-D) flows are nonnegative. This assumption is typically not problematic for small- and medium-scale networks in which the EKF has been successfully applied. However, in large-scale networks (which typically contain numbers of O-D pairs with small magnitudes of flow), the estimates may severely violate constraints. In consequence, simply truncating the infeasible estimates may result in the divergence of EKF, leading to extremely poor state estimations and predictions. To address this issue, a constrained EKF (CEKF) approach is presented; it imposes constraints on the posterior distribution of the state estimators to obtain the maximum a posteriori (MAP) estimates that are feasible. The MAP estimates are obtained with a heuristic followed by the coordinate descent method. The procedure determines the optimum and are computationally faster by 31.5% over coordinate descent and by 94.9% over the interior point method. Experiments on the Singapore expressway network indicated that the CEKF significantly improved model accuracy and outperformed the traditional EKF (up to 78.17%) and generalized least squares (up to 17.13%) approaches in state estimation and prediction.

On point estimation of the abnormality of a Mahalanobis index

Mahalanobis distance may be used as a measure of the disparity between an individual’s profile of scores and the average profile of a population of controls. The degree to which the individual’s profile is unusual can then be equated to the proportion of the population who would have a larger Mahalanobis distance than the individual. Several estimators of this proportion are examined. These include plug-in maximum likelihood estimators, medians, the posterior mean from a Bayesian probability matching prior, an estimator derived from a Taylor expansion, and two forms of polynomial approximation, one based on Bernstein polynomial and one on a quadrature method. Simulations show that some estimators, including the commonly-used plug-in maximum likelihood estimators, can have substantial bias for small or moderate sample sizes. The polynomial approximations yield estimators that have low bias, with the quadrature method marginally to be preferred over Bernstein polynomials. However, the polynomial estimators sometimes yield infeasible estimates that are outside the 0–1 range. While none of the estimators are perfectly unbiased, the median estimators match their definition; in simulations their estimates of the proportion have a median error close to zero. The standard median estimator can give unrealistically small estimates (including 0) and an adjustment is proposed that ensures estimates are always credible. This latter estimator has much to recommend it when unbiasedness is not of paramount importance, while the quadrature method is recommended when bias is the dominant issue.

Interval shrinkage estimators

This paper considers estimation of an unknown distribution parameter in situations where we believe that the parameter belongs to a finite interval. We propose for such situations an interval shrinkage approach which combines in a coherent way an unbiased conventional estimator and non-sample information about the range of plausible parameter values. The approach is based on an infeasible interval shrinkage estimator which uniformly dominates the underlying conventional estimator with respect to the mean square error criterion. This infeasible estimator allows us to obtain useful feasible counterparts. The properties of these feasible interval shrinkage estimators are illustrated both in a simulation study and in empirical examples.

Interval Shrinkage Estimators

This paper considers estimation of an unknown distribution parameter in situations where we believe that the parameter belongs to a finite interval. We propose for such situations an interval shrinkage approach which combines in a coherent way an unbiased conventional estimator and non-sample information about the range of plausible parameter values. The approach is based upon an infeasible interval shrinkage estimator which uniformly dominates the underlying conventional estimator with respect to the mean square error criterion. This infeasible estimator allows us to obtain useful feasible counterparts. The properties of these feasible interval shrinkage estimators are illustrated both in a simulation study and in empirical examples.

Pseudo-maximum likelihood estimation in two classes of semiparametric diffusion models

Two classes of semiparametric diffusion models are considered, where either the drift or the diffusion term is parameterized, while the other term is left unspecified. We propose a pseudo-maximum likelihood estimator (PMLE) of the parametric component that maximizes the likelihood with a preliminary estimator of the unspecified term plugged in. It is demonstrated how models and estimators can be used in a two-step specification testing strategy of semiparametric and fully parametric models, and shown that approximate/simulated versions of the PMLE inherit the properties of the actual but infeasible estimator. A simulation study investigates the finite sample performance of the PMLE.

Pseudo-Maximum Likelihood Estimation in Two Classes of Semiparametric Diffusion Models

A novel estimation method for two classes of semiparametric scalar diffusion models is proposed: In the first class, the diffusion term is parameterised and the drift is left unspecified, while in the second class only the drift term is specified. Under the assumption of stationarity, the unspecified term can be identified as a functional of the parametric component and the stationary density. Given a discrete sample with a fixed time distance, the parametric component is then estimated by maximizing the associated likelihood with a preliminary estimator of the unspecified term plugged in. It is shown that this Pseudo-MLE (PMLE) is root n-consistent and asymptotically normally distributed under regularity conditions, and demonstrate how the models and estimators can be used in a two-step specification testing strategy of fully parametric models. Since the likelihood function is not available on closed form, the practical implementation of our estimator and tests will rely on simulated or approximate PMLE's. Under regularity conditions, it is verified that approximate/simulated versions of the PMLE inherits the properties of the actual but infeasible estimator. A simulation study investigates the finite-sample performance of the PMLE, and finds that it performs well and is comparable to parametric MLE both in terms of bias and variance.

Spline confidence bands for variance functions

Asymptotically exact and conservative confidence bands are obtained for possibly heteroscedastic variance functions, using piecewise constant and piecewise linear spline estimation, respectively. The variance estimation is as efficient as an infeasible estimator when the conditional mean function is known, and the widths of the confidence bands are of optimal order. Simulation experiments provide strong evidence that corroborates the asymptotic theory while the computing is extremely fast. A slower bootstrap band is also proposed, with much higher accuracy. As illustrations, the bootstrap spline band has been applied to test for heteroscedasticity in fossil data and in motorcycle data.

A NON-HOMOGENEOUS REGULARIZATION METHOD FOR THE ESTIMATION OF NARROW AEROSOL SIZE DISTRIBUTIONS

The determination of particle size distributions from e.g. DMPS or TDMA measurements is an ill-posed problem that is often solved using regularization methods. With Tikhonov-type regularization the standard smoothness assumptions may yield infeasible estimates for the size distribution in cases where the size distribution contains narrow peaks. In this paper we propose a method for the estimation of size distributions with narrow peaks. The method is based on the utilization of a weight function to modify the standard smoothness constraints to obtain non-homogeneous smoothing effect. In the problem formulation prior assumptions about the weight function are exploited and the size distribution and the weight function can be estimated simultaneously. The performance of the method is evaluated with simulated TDMA data and the results show that good agreement between the estimates and the true size distributions can be achieved.