Asymptotic relative efficiency of the linear discriminant function under partial nonrandom classification of the training data

This paper considers the problem where the linear discriminant rule is formed from training data that are only partially classified with respect to the two groups of origin. A further complication is that the data of unknown origin do not constitute an observed random sample from a mixture of the two under- lying groups. Under the assumption of a homoscedastic normal model, the overall error rate of the sample linear discriminant rule formed by maximum likelihood from the partially classified training data is derived up to and including terms of the first order in the case of univariate feature data. This first- order expansion of the sample rule so formed is used to define its asymptotic efficiency relative to the rule formed from a completely classified random training set and also to the rule formed from a completely unclassified random set.     

Computing optimal adjustment schemes for the general tool-wear problem

An important problem in process adjustment using feedback is how often to sample the process and when and by how much to apply an adjustment. Minimum cost feedback schemes based on simple, but practically interesting, models for disturbances and dynamics have been discussed in several particular cases. The more general situation in which there may be measurement and adjustment errors, deterministic process drift, and costs of taking an observation, of making an adjustment, and of being off target, is considered in this article. Assuming all these costs to be known, a numerical method to minimize the overall expected cost is presented. This numerical method provides the optimal sampling interval, action limits, and amount of adjustment; and the resulting average adjustment interval, mean squared deviation from target, and minimum overall expected cost. When the costs of taking an observation, of making an adjustment, and of being off target are not known, the method can be used to choose a particular scheme by judging the advantages and disadvantages of alternative options considering the mean squared deviation they produce, the frequency with which they require observations to be made, and the resulting overall length of time between adjustments. Computer codes that perform the required computations are provided in the appendices and applied to find optimal adjustment schemes in three real examples of application.     

Bootstrapping estimators for the seemingly unrelated regressions model

In this paper we consider the possibility of using the bootstrap to estimate the finite sample variability of feasible generalized least squares and improved estimators applied to the seemingly unrelated regressions model. The improved estimators we employ include members of the Stein-rule family and a hierarchical Bayes estimator proposed by Blattberg and George (1991). Simulation experiments are carried out using several SUR examples as well as a very large example based on the price-promotion model, and data, from marketing research.     

Statistical Estimates for Modified Gravitation Model by Aggregated Data

A nonlinear regression model for forecasting of passenger flow between various spatial points (towns) is described. Unknown parameters are estimated using aggregated data when the information about a number of the departed passengers from each town is available only. For estimation, the least squares and maximum likelihood methods are used. Numerical examples are performed to illustrate the proposed approaches.     

Machine Learning Vasicek Model Calibration with Gaussian Processes

In this article, we calibrate the Vasicek interest rate model under the risk neutral measure by learning the model parameters using Gaussian processes for machine learning regression. The calibration is done by maximizing the likelihood of zero coupon bond log prices, using mean and covariance functions computed analytically, as well as likelihood derivatives with respect to the parameters. The maximization method used is the conjugate gradients. The only prices needed for calibration are zero coupon bond prices and the parameters are directly obtained in the arbitrage free risk neutral measure.     

Linear Models with Doubly Exchangeable Distributed Errors

We study the general linear model (GLM) with doubly exchangeable distributed error for m observed random variables. The doubly exchangeable general linear model (DEGLM) arises when the m-dimensional error vectors are “doubly exchangeable,” jointly normally distributed, which is a much weaker assumption than the independent and identically distributed error vectors as in the case of GLM or classical GLM (CGLM). We estimate the parameters in the model and also find their distributions. We show that the tests of intercept and slope are possible in DEGLM as a particular case using parametric bootstrap as well as multivariate Satterthwaite approximation.     

Soft Classification Trees

An extension of soft classification trees to multinomial outcomes is presented. Estimates of the method's predictive accuracy, as well as average tree size and tree depths, are systematically compared to those of the conventional Classification and Regression Tree (CART) approach by the means of simulations. A similar comparison is performed on real datasets. Results point to an advantage in favor of the soft tree.     

Estimation of the Mean of a Sensitive Variable in the Presence of Auxiliary Information

Sousa et al. (2010 Sousa , R. , Shabbir , J. , Real , P. C. , Gupta , S. ( 2010 ). Ratio estimation of the mean of a sensitive variable in the presence of auxiliary information . J. Statist. Theor. Prac. 4 ( 3 ): 495 – 507 .[Taylor & Francis Online] , [Google Scholar]) introduced a ratio estimator for the mean of a sensitive variable and showed that this estimator performs better than the ordinary mean estimator based on a randomized response technique (RRT). In this article, we introduce a regression estimator that performs better than the ratio estimator even for modest correlation between the primary and the auxiliary variables. The underlying assumption is that the primary variable is sensitive in nature but a non sensitive auxiliary variable exists that is positively correlated with the primary variable. Expressions for the Bias and MSE (Mean Square Error) are derived based on the first order of approximation. It is shown that the proposed regression estimator performs better than the ratio estimator and the ordinary RRT mean estimator (that does not utilize the auxiliary information). We also consider a generalized regression-cum-ratio estimator that has even smaller MSE. An extensive simulation study is presented to evaluate the performances of the proposed estimators in relation to other estimators in the study. The procedure is also applied to some financial data: purchase orders (a sensitive variable) and gross turnover (a non sensitive variable) in 2009 for a population of 5,336 companies in Portugal from a survey on Information and Communication Technologies (ICT) usage.     

A Small Area Predictor under Area-Level Linear Mixed Models with Restrictions

A calibrated small area predictor based on an area-level linear mixed model with restrictions is proposed. It is showed that such restricted predictor, which guarantees the concordance between the small area estimates and a known estimate at the aggregate level, is the best linear unbiased predictor. The mean squared prediction error of the calibrated predictor is discussed. Further, a restricted predictor under a particular time-series and cross-sectional model is presented. Within a simulation study based on real data collected from a longitudinal survey conducted by a national statistical office, the proposed estimator is compared with other competitive restricted and non-restricted predictors.     

An Improved Estimation in Regression Parameter Matrix in Multivariate Regression Model

We consider shrinkage and preliminary test estimation strategies for the matrix of regression parameters in multivariate multiple regression model in the presence of a natural linear constraint. We suggest a shrinkage and preliminary test estimation strategies for the parameter matrix. The goal of this article is to critically examine the relative performances of these estimators in the direction of the subspace and candidate subspace restricted type estimators. Our analytical and numerical results show that the proposed shrinkage and preliminary test estimators perform better than the benchmark estimator under candidate subspace and beyond. The methods are also applied on a real data set for illustrative purposes.