摘要:

A new branch and bound algorithm is described for solving non-negative least squares problems. The method is extended to the case where a single equality constraint must also be satisfied. Simulations suggest that the new algorithm is considerably faster than several alternative methods, especially when most of the parameter estimates are strictly positive at the solution.

ISSN:0361-0918    

DOI:10.1080/03610919308813095

出版商:Marcel Dekker, Inc.    

年代:1993

数据来源: Taylor

摘要:

In many instances a fixed number of items, N, must be obtained from a large collection of these items. The process of counting out these items, however, becomes impractical if N is quite large. An alternative to individually counting out N items is counting by weighing. If the mean weight of an individual item, μ, is known, then we simply assemble a batch that weighs Nμ. If the mean weight is unknown, then we take an initial sample of size n, much less than N, from which an estimate, m, of the mean weight is obtained. We then assemble a batch that weighs (N−n)m.This procedure leads in principle to a set of N total items (n counted, N−n weighed). By way of renewal theory, this articleexamines the distributional properties of the actual number of items in the batch. Further, from the distributional properties of the actual number ofitems counted, this, article addresses the problem of determining the smallest initial sample size n for estimating N to within some specified bound with high probability. Also, refinements known as overshoot and continuityw corrections are implemented to improve the procedure. Finally, a simulation study was performed to evaluate the performance of the procedure.

ISSN:0361-0918    

DOI:10.1080/03610919308813096

出版商:Marcel Dekker, Inc.    

年代:1993

数据来源: Taylor

摘要:

Bolfarine 1991 has considered a location measurement error superpopulation model and obtained Bayes predictors of the population total and population variance under noninformative priors. In this paper, those results are first extended to the case of informative priors. Next, empirical Bayes predictors of the two population quantities are considered using the approach which has been proposed by Ghosh and Meeden in the context of repeated surveys and small area estimation. Extensions are considered to the "nonho-mogeneous" superpopulation model considered by Tiwari and Lahiri (1989). Some simulations studies which illustrate the effect of the measurement error in the proposed empirical Bayes predictors are also reported.

ISSN:0361-0918    

DOI:10.1080/03610919308813097

出版商:Marcel Dekker, Inc.    

年代:1993

数据来源: Taylor

摘要:

A single variable sampling plan with a polynomial loss function is studied. The variable measured from each item of a 2 batch follows a normal distribution N(m,t2) where m has a normal prior distribution, and t2has either a uniform prior distribution or a two-parameter exponential prior distribution. An algorithm for the determination of an optimal sampling planis given. The algorithm isfinite, i.e. an optimal solution can be obtained after a finite number of steps of searching. 371 Copyright 1993 by Marcel Dekker, Inc

ISSN:0361-0918    

DOI:10.1080/03610919308813098

出版商:Marcel Dekker, Inc.    

年代:1993

数据来源: Taylor

摘要:

A multiparameter extension is made to modified two stage shrinkage estimator proposed by Handa andKambo (1990), For aparticular class of shrinkage estimator, the local optimality of the extended modified estimator is shown over the two stage shrinkage estimator defined by Bhattacharya and Prakasa Rao(1990) in terms of quadratic loss.

ISSN:0361-0918    

DOI:10.1080/03610919308813099

出版商:Marcel Dekker, Inc.    

年代:1993

数据来源: Taylor

摘要:

Negative Binomial Distribution (NBD) has been widely employed in marketing to model purchase incidence. In this paper, we undertake a simulation study to examine the sensitivity of NBD when the number of purchases in a given period is Poisson or Erlang 2 but the underlying heterogeneity is different from gamma. In particular, we allow the heterogeneity to follow three other flexible and well-known distributions: inverse Gaussian, lognormal and Weibull. All these distributions have shapes that are similar to the gamma. Moreover, we evaluate the performance of three alternative models namely, the Condensed Negative Binomial (CNBD), the Generalized Negative Binomial (GNBD) and the Generalized Poisson (GPD).

ISSN:0361-0918    

DOI:10.1080/03610919308813100

出版商:Marcel Dekker, Inc.    

年代:1993

数据来源: Taylor

摘要:

Critical values for Dunnett's multiple comparison procedure for simultaneously comparing the means of several experimental treatments with the mean of a control treatment are available in many references. However, these values are primarily applicable to one-way analysis of variance with equal sample sizes for the experimental treatments. Even for this design, these critical values do not suffice for situations where the corresponding p-values are desired. Here the technique is presented for calculation of the p-value for multiple comparisons with a control when a general linear model is used for the analysis. A FORTRAN program using IMSL subroutines is described for the calculations.

ISSN:0361-0918    

DOI:10.1080/03610919308813101

出版商:Marcel Dekker, Inc.    

年代:1993

数据来源: Taylor

摘要:

While many tests of univariate and multivariate normality have been proposed, those based on skewness and kurtosis coefficients are widely presumed to offer the advantage of diagnosing how distributions depart from normality. However, results summarized from many Monte Carlo studies show that tests based on skewness coefficients do not reliably discriminate between skewed and non-skewed distributions. Indeed, the use of skewness tests to discriminate between these distributions lackstheoretical foundation. The performance of skewness tests is shown to be very sensitive to the kurtosis of the underlying distribution

ISSN:0361-0918    

DOI:10.1080/03610919308813102

出版商:Marcel Dekker, Inc.    

年代:1993

数据来源: Taylor

摘要:

The Qosand Qm are two leading estimators of the probability of misclassification which are based on the asymptotic expansion of the the expected value of the Error Rate, Pi. The estimators are, however, not suitable for estimating the Error rates for certain ranges of the parameters p , n1, n2and ß.We investigate the regions in which they produce unacceptable estimates , and show that the Qosis, in general, better than the Qmin producing acceptable estimates

ISSN:0361-0918    

DOI:10.1080/03610919308813103

出版商:Marcel Dekker, Inc.    

年代:1993

数据来源: Taylor

摘要:

In this paper four classical and four Bayes estimators of the exponential reliability function are compared under time censored sampling with replacement. The classical estimators studied are the maximum likelihood and minimum variance unbiasedand two so-called shrunken estimators, the K and a estimators. Two kinds of Bayes estimators, for each of the gamma and inverted gamma prior distributions for exponential parameter, are studied. Comparisons are made based on mean square errors,and it is shown that no classical estimator is best over the entire parameter space. In a special, though artificial sense,the Bayes estimators for gamma priors perform generally best among all estimators

ISSN:0361-0918    

DOI:10.1080/03610919308813104

出版商:Marcel Dekker, Inc.    

年代:1993

数据来源: Taylor