Building a statistical model to predict reactor temperatures

This paper describes the various stages in building a statistical model to predict temperatures in the core of a reactor, and compares the benefits of this model with those of a physical model. We give a brief background to this study and the applications of the model to rapid online monitoring and safe operation of the reactor. We describe the methods, of correlation and two dimensional spectral analysis, which we use to identify the effects that are incorporated in a spatial regression model for the measured temperatures. These effects are related to the age of the reactor fuel and the spatial geometry of the reactor. A remaining component of the temperature variation is a slowly varying temperature surface modelled by smooth functions with constrained coefficients. We assess the accuracy of the model for interpolating temperatures throughout the reactor, when measurements are available only at a reduced set of spatial locations, as is the case in most reactors. Further possible improvements to the model are discussed.     

An m-estimation-based model selection criterion with a data-oriented penalty

In Wu and Zen (1999), a linear model selection procedure based on M-estimation is proposed, which includes many classical model selection criteria as its special cases, and it is shown that the selection procedure is strongly consistent for a variety of penalty functions. In this paper, we will investigate its small sample performances for some choices of fixed penalty functions. It can be seen that the performance varies with the choice of the penalty. Hence, a randomized penalty based on observed data is proposed, which preserves the consistency property and provides improved performance over a fixed choice of penalty functions.     

Coversge probabilty to the original data by prediction intervals in simple linear regression

The coverage rate of the original data by the prediction interval in simple linear regression is obtained by computer simulation. The results show that for small sample size, the coverage rate is higher than the assigned prediction coverage rate (confidence level). The two coverage rates begin to converge when the sample size is larger than 50 and the convergence rate depends very little on the distribution of the independent variable. Also, theoretical results on the asymptotic coverage rate and on the absolute minimum bounds are obtained     

基于半参数空间模型的空气污染与经济增长关系再检验

选取2001-2014年中国30个省份数据作为样本,考虑空气污染的空间自相关性,采用空间杜宾滞后模型（SDM）和半参数空间滞后模型实证检验经济发展与空气污染的非线性关系。结果表明：①中国空气污染存在显著的空间正相关性,高空气污染水平集聚地随时间推移呈现出“由西向东”的转移特征。②空气污染与经济增长存在一种震荡曲线形式的关系,并不完全吻合传统的EKC倒U型曲线形状,但震荡关系也符合EKC所描述的环境污染与经济发展的关系将长期存在的特征,说明了经济增长并不能自发解决空气污染问题。③半参数空间滞后模型的拟合优度高于普通参数模型,其刻画的空气污染与经济增长的非线性特征验证了前人对二者震荡关系的猜想,结果更为稳健、准确与有效。     

Simplified hierarchical linear models for the evaluation of surrogate endpoints

The linear mixed-effects model (Verbeke and Molenberghs, 2000) has become a standard tool for the analysis of continuous hierarchical data such as, for example, repeated measures or data from meta-analyses. However, in certain situations the model does pose insurmountable computational problems. Precisely this has been the experience of Buyse et al. (2000a) who proposed an estimation- and prediction-based approach for evaluating surrogate endpoints. Their approach requires fitting linear mixed models to data from several clinical trials. In doing so, these authors built on the earlier, single-trial based, work by Prentice (1989), Freedman et al. (1992), and Buyse and Molenberghs (1998). While Buyse et al. (2000a) claim their approach has a number of advantages over the classical single-trial methods, a solution needs to be found for the computational complexity of the corresponding linear mixed model. In this paper, we propose and study a number of possible simplifications. This is done by means of a simulation study and by applying the various strategies to data from three clinical studies: Pharmacological Therapy for Macular Degeneration Study Group (1977), Ovarian Cancer Meta-analysis Project (1991) and Corfu-A Study Group (1995).     

Some New Estimators for Small-Area Means With Application to the Assessment of Farmland Values

Regression models that account for main state effects and nested county effects are considered for the assessment of farmland values. Empirical predictors obtained by replacing the unknown variances in the formulas of the optimal predictors by maximum likelihood estimates are presented. The computations are carried out by simple iterations between two SAS procedures. Estimators for the prediction variances are derived, and a modification to secure the robustness of the predictors is proposed. The procedure is applied to data on nonirrigated cropland in the Corn Belt states and is shown to yield predictors with considerably lower prediction mean squared errors than the survey estimators and other regression-type estimators.     

An alternate version of the conceptual predictive statistic based on a symmetrized discrepancy measure

The conceptual predictive statistic, C_p, is a widely used criterion for model selection in linear regression. C_p serves as an estimator of a discrepancy, a measure that reflects the disparity between the generating model and a fitted candidate model. This discrepancy, based on scaled squared error loss, is asymmetric: an alternate measure is obtained by reversing the roles of the two models in the definition of the measure. We propose a variant of the C_p statistic based on estimating a symmetrized version of the discrepancy targeted by C_p. We claim that the resulting criterion provides better protection against overfitting than C_p, since the symmetric discrepancy is more sensitive towards detecting overspecification than its asymmetric counterpart. We illustrate our claim by presenting simulation results. Finally, we demonstrate the practical utility of the new criterion by discussing a modeling application based on data collected in a cardiac rehabilitation program at University of Iowa Hospitals and Clinics.     

Confidence Intervals Following Group Sequential Tests in Clinical Trails with Multivariate Observations

Several authors ( e.g. Kim and DeMets, 1987a, Biometrics) have developed methods for estimation following group sequential tests in clinical trials when each patient has only one response. In many long-term clinical trials, the subjects enter the study sequentially and yield repeated measurements or other types of multivariate observations at successive follow-up visits. Typically, investigators want to compare a parameter of interest such as the slope over time in a repeated measures trial etc. In this article, we propose an exact confidence interval for these parameters in a repeated measures trial, and compare it with a naive confidence interval using Monte Carlo simulation. This method is illustrated with a real example for bone density measurements.     

Applying discrete choice models to predict Academy Award winners

Summary.  Every year since 1928, the Academy of Motion Picture Arts and Sciences has recognized outstanding achievement in film with their prestigious Academy Award, or Oscar. Before the winners in various categories are announced, there is intense media and public interest in predicting who will come away from the awards ceremony with an Oscar statuette. There are no end of theories about which nominees are most likely to win, yet despite this there continue to be major surprises when the winners are announced. The paper frames the question of predicting the four major awards—picture, director, actor in a leading role and actress in a leading role—as a discrete choice problem. It is then possible to predict the winners in these four categories with a reasonable degree of success. The analysis also reveals which past results might be considered truly surprising—nominees with low estimated probability of winning who have overcome nominees who were strongly favoured to win.     

Bayesian predictive analysis of the linear regression model with an edgeworth series prior distribution

In this paper we obtain the Bayes forecasts for the future observations on the dependent variable in the linear regression model when the regression coefficients have an Edgeworth series prior distribution. Furthermore, we consider the effect of departure from normality of the prior distribution of regression coefficients on the Bayes forecasts.     

区域差异导致经济块状发展的空间机理研究

 本文运用空间计量经济学模型以及修正绝对β收敛回归模型,对中国2003—2007年省域经济发展的集聚性与差异性进行研究发现：中国31个省域之间存在着较强的空间集聚和空间依赖性,虽受到地域限制,但集聚辐射区域有逐步扩大的趋势;固定资产投入、人均实际工资以及城镇化率对于区域经济集聚的发展起到稳健的推动作用,FDI对集聚区域发展的促进作用也越加显著,但是财政投入却在一定层度上遏制了块状经济的发展。另外,修正和传统收敛模型均显示,2003-2007年我国经济整体发展呈现明显的不收敛特性。     

Some factors affecting statistical power of approximate tests in the linear mixed model for longitudinal data

Different longitudinal study designs require different statistical analysis methods and different methods of sample size determination. Statistical power analysis is a flexible approach to sample size determination for longitudinal studies. However, different power analyses are required for different statistical tests which arises from the difference between different statistical methods. In this paper, the simulation-based power calculations of F-tests with Containment, Kenward-Roger or Satterthwaite approximation of degrees of freedom are examined for sample size determination in the context of a special case of linear mixed models (LMMs), which is frequently used in the analysis of longitudinal data. Essentially, the roles of some factors, such as variance–covariance structure of random effects [unstructured UN or factor analytic FA0], autocorrelation structure among errors over time [independent IND, first-order autoregressive AR1 or first-order moving average MA1], parameter estimation methods [maximum likelihood ML and restricted maximum likelihood REML] and iterative algorithms [ridge-stabilized Newton-Raphson and Quasi-Newton] on statistical power of approximate F-tests in the LMM are examined together, which has not been considered previously. The greatest factor affecting statistical power is found to be the variance–covariance structure of random effects in the LMM. It appears that the simulation-based analysis in this study gives an interesting insight into statistical power of approximate F-tests for fixed effects in LMMs for longitudinal data.     

Empirical bayes approach to parameter identification in linear stochastic difference equations

Empirical Bayesian parameter estimators and predictors for linear stochastic difference equations are constructed and discussed. Some properties as consistency and asymptotic optimality are investigated. The given methods are illustrated by the example of a univariate first order autoregressive process.     

A Bayesian statistical model for end member analysis of sediment geochemistry, incorporating spatial dependences

Summary.  An important problem in the management of water supplies is identifying the sources of sediment. The paper develops a Bayesian approach, utilizing an end member model, to estimate the proportion of various sources of sediments in samples taken from a dam. This approach not only allows for the incorporation of prior knowledge about the geochemical compositions of the sources (or end members) but also allows for correlation between spatially contiguous samples and the prediction of the sediment's composition at unsampled locations. Sediments that were sampled from the North Pine Dam in south-east Queensland, Australia, are analysed to illustrate the approach.     

The inverse binomial distribution as a statistical model

A particular case of Jain and Consul's (1971) generalized neg-ative binomial distribution is studied. The name inverse binomial is suggested because of its close relation with the inverse Gaussian distribution. We develop statistical properties including conditional inference of a parameter. An application using real data is given.     

C472. The expected value of a conditional variance: An upper bound

The article derives Bartlett corrections for improving the chi-square approximation to the likelihood ratio statistics in a class of symmetric nonlinear regression models. This is a wide class of models which encompasses the t model and several other symmetric distributions with longer-than normal tails. In this paper we present, in matrix notation, Bartlett corrections to likelihood ratio statistics in nonlinear regression models with errors that follow a symmetric distribution. We generalize the results obtained by Ferrari, S. L. P. and Arellano-Valle, R. B. (1996). Modified likelihood ratio and score tests in linear regression models using the t distribution. Braz. J. Prob. Statist., 10, 15–33, who considered a t distribution for the errors, and by Ferrari, S. L. P. and Uribe-Opazo, M. A. (2001). Corrected likelihood ratio tests in a class of symmetric linear regression models. Braz. J. Prob. Statist., 15, 49–67, who considered a symmetric linear regression model. The formulae derived are simple enough to be used analytically to obtain several Bartlett corrections in a variety of important models. We also present simulation results comparing the sizes and powers of the usual likelihood ratio tests and their Bartlett corrected versions.     

A non-iterative approach to estimating parameters in a linear structural equation model

The research described herein was motivated by a study of the relationship between the performance of students in senior high schools and at universities in China. A special linear structural equation model is established, in which some parameters are known and both the responses and the covariables are measured with errors. To explore the relationship between the true responses and latent covariables and to estimate the parameters, we suggest a non-iterative estimation approach that can account for the external dependence between the true responses and latent covariables. This approach can also deal with the collinearity problem because the use of dimension-reduction techniques can remove redundant variables. Combining further with the information that some of parameters are given, we can perform estimation for the other unknown parameters. An easily implemented algorithm is provided. A simulation is carried out to provide evidence of the performance of the approach and to compare it with existing methods. The approach is applied to the education example for illustration, and it can be readily extended to more general models.