Group variable selection in cardiopulmonary cerebral resuscitation data for veterinary patients

Cardiopulmonary cerebral resuscitation (CPCR) is a procedure to restore spontaneous circulation in patients with cardiopulmonary arrest (CPA). While animals with CPA generally have a lower success rate of CPCR than people do, CPCR studies in veterinary patients have been limited. In this paper, we construct a model for predicting success or failure of CPCR, and identifying and evaluating factors that affect the success of CPCR in veterinary patients. Due to reparametrization using multiple dummy variables or close proximity in nature, many variables in the data form groups, and thus a desirable method should take this grouping feature into account in variable selection. To accomplish these goals, we propose an adaptive group bridge method for a logistic regression model. The performance of the proposed method is evaluated under different simulated setups and compared with several other regression methods. Using the logistic group bridge model, we analyze data from a CPCR study for veterinary patients and discuss their implications on the practice of veterinary medicine.     

On Bayesian learning via loss functions

We provide a decision theoretic approach to the construction of a learning process in the presence of independent and identically distributed observations. Starting with a probability measure representing beliefs about a key parameter, the approach allows the measure to be updated via the solution to a well defined decision problem. While the learning process encompasses the Bayesian approach, a necessary asymptotic consideration then actually implies the Bayesian learning process is best. This conclusion is due to the requirement of posterior consistency for all models and of having standardized losses between probability distributions. This is shown considering a specific continuous model and a very general class of discrete models.     

Conditional logspline density estimation

In conditional logspline modelling, the logarithm of the conditional density function, log f(y|x), is modelled by using polynomial splines and their tensor products. The parameters of the model (coefficients of the spline functions) are estimated by maximizing the conditional log-likelihood function. The resulting estimate is a density function (positive and integrating to one) and is twice continuously differentiable. The estimate is used further to obtain estimates of regression and quantile functions in a natural way. An automatic procedure for selecting the number of knots and knot locations based on minimizing a variant of the AIC is developed. An example with real data is given. Finally, extensions and further applications of conditional logspline models are discussed.     

Calibration of the estimators of variance

This investigation suggests new techniques to calibrate estimators of variance. Estimators of the variance of simple mean, ratio and regression estimators under different sampling schemes are shown to be special cases of the proposed calibration techniques. The approach has more practical use due to recent advances in programming techniques and computational speed. An empirical study has been carried out to address the properties of these proposed strategies.     

On the Sequential Selection Problem*

Kwan and Yuan [13] considered the sequential selection problem in which an employer should arrange the sequence of interviews with job applicants to fill a position. In this note, it is shown that their selection problem would be alternately interpreted as an optimal search problem or, more specifically, a discrete search problem with a stationary target. A more generalized version of the ordering problem is proposed which explicitly considers the time value of money. Also, the optimality of the ordering strategy in the generalized problem is proven by the pair-wise exchange method, which is simpler than the induction hypothesis-based proof. The sequention selection problem is shown to be a special case of the general ordering problem where the discount rate is zero.     

A Comparative Evaluation of Labor Tour Scheduling Methods

This paper presents an initial study of relative performance for a number of the labor tour scheduling heuristic methods proposed in the literature. These heuristic methods were classified as either linear programming (LP) based or construction. Each of the methods was applied to a tour scheduling problem, subject to a variety of labor demand requirements distributions, with the singular objective being the minimization of total labor hours scheduled. Statistical analysis revealed that effective tour schedule solutions were generated by both LP-based and construction methods. Since the performances of the Keith [13], Morris and Showalter [18], and Bechtold and Showalter [5] methods were superior, their solutions were also compared across a number of secondary criteria. An overall analysis of the performances of these three methods resulted in the identification of a number of important managerial and decision-making issues. We conclude that service operations management should consider integrating these heuristic methods into a decision support system. Finally, suggestions for future research are provided.