ISSN:0040-1706    

DOI:10.1080/00401706.1991.10484865

出版商:Taylor & Francis Group    

年代:1991

数据来源: Taylor

ISSN:0040-1706    

DOI:10.1080/00401706.1991.10484866

出版商:Taylor & Francis Group    

年代:1991

数据来源: Taylor

摘要:

Classical experimental design is based on the three concepts of randomization, blocking, and replication. Randomization endeavors to neutralize the effects of (spatial) correlation and yields valid tests for the hypothesis of equal treatment effects. More recently, attempts have been made to use the spatial location of treatments to improve the efficiencies of estimators of treatment contrasts. In this article, we show that a simple, flexible spatial-modeling approach to the analysis of industrial experiments (e.g., wafer fabrication) can yield more efficient estimators of the treatment contrasts than the classical approach. We base the analysis on empirical generalized least squares estimation, in which the spatial-dependence parameters are estimated from resistantly detrended response data.

ISSN:0040-1706    

DOI:10.1080/00401706.1991.10484867

出版商:Taylor & Francis Group    

年代:1991

数据来源: Taylor

摘要:

We investigate the performance of Dodge's continuous sampling plan 1 (CSP-1) when production run lengths are short or moderate or when the input process is not iid Bernoulli. For finite run lengths. methods are presented for computing upper and lower percentage points (probability limits) for the distributions of the outgoing quality and fraction inspected when the input process is Markov. Both rectifying and nonrectifying inspection are considered. Particular attention is paid to the special case of iid Bernoulli input. Comparisons are made with long-run maximum averages under various input process assumptions, including Dodge's average outgoing quality limit (AOQL) for the iid case, an AOQL for Markov input, and unrestricted AOQL values, These limits arc proposed as new performance measures. intended to supplement the tables in MIL-STD-1235B for use in selecting CSP-1 plans. We also discuss the problem of interval estimation of the incoming and outgoing quality from CSP-1 inspection data.

ISSN:0040-1706    

DOI:10.1080/00401706.1991.10484868

出版商:Taylor & Francis Group    

年代:1991

数据来源: Taylor

摘要:

A set ofqresponses,y= (y1,y2, …,yq,)T, is related to a set ofpexplanatory variables,x= (x1,x2, …,xp)T, through the classical linear regression model,yT=aT+xTB+eT. First, the unknown parametersaandBare estimated using a calibrafion set. The statistical problem that is considered here is that of estimating the vectorxo, that underlies a new observed vector of responsesyousing the parameter estimates obtained from the first procedure. These two procedures are commonly referred to ascalibration and prediction(orinverse prediction) and sometimes jointly referred to as calibration. The prediction procedure can be viewed as parameter estimation in errors-in-variables regression. Themaximum likelihood estimator(assuming normally distributed measurement errors) is proposed for the prediction procedure. Unlike the classical estimator used in the prediction procedure, the proposed estimator is consistent with respect to the number of response variables. The performances of the maximum likelihood estimator and the classical estimator are compared both analytically and via Monte Carlo simulations. An example is given frominfrared spectroscopy.

ISSN:0040-1706    

DOI:10.1080/00401706.1991.10484869

出版商:Taylor & Francis Group    

年代:1991

数据来源: Taylor

摘要:

Taguchi's robust-design technique, also known as parameter design, focuses on making product and process designs insensitive (i.e., robust) to hard-to-control variations. In some applications, however, his approach of modeling expected loss and the resulting “product array” experimental format leads to unnecessarily expensive experiments. As an alternative to Taguchi's “loss model/product array” formulation. Welch, Yu, Kang, and Sacks proposed combining control and noise factors in a single array, modeling the response itself rather than expected loss, and then approximating a prediction model for loss based on the fitted-response model. In this article, we further develop and strengthen this response-model/combined-array approach. We recommended examination of control-by-noise interaction plots suggested by the fitted-response model. These plots can reveal control-factor settings that dampen the effects of individual noise factors. We also show that the run savings from using combined arrays are due to the flexibility that this formulation allows for estimation of effects. In contrast, the product-array formulation dictates estimation of many effects that are unlikely to be important.

ISSN:0040-1706    

DOI:10.1080/00401706.1991.10484870

出版商:Taylor & Francis Group    

年代:1991

数据来源: Taylor

摘要:

We study in this article the energies of neutrinos observed in the Irvine-Michigan-Brookhaven detector near Fairport, Ohio. The chance of observing a neutrino with a given amount of energy is a function of the sensitivity ortrigger efficiencyof the detector. Because some of the neutrinos were undetected, the distribution of the observed data is randomly truncated. We consider maximum likelihood estimation of the parameters for the randomly truncated normal distribution (as well as a normal model for power transformations). To obtain confidence regions for the parameters, both asymptotic-normal-theory and approximate-likelihood-based confidence regions are constructed. Monte Carlo investigations are used to study the properties of maximum likelihood estimators for randomly truncated normal distributions. The simulation study shows that confidence regions based on the asymptotic normal theory perform very poorly. whereas the approximate-likelihood-based regions provide coverages closer to the nominal level.

ISSN:0040-1706    

DOI:10.1080/00401706.1991.10484871

出版商:Taylor & Francis Group    

年代:1991

数据来源: Taylor

摘要:

When measurements are made on units of production in time order, a problem of considerable importance is to test whether the random disturbances from a regression model are autocorrelated in time. Tool-wear processes provide an example that is important in itself and for which regression models can sometimes be useful in modeling and controlling the process. For these processes, because of compensator actions and direct interventions inter alia, the deterministic components of the data are frequently from regression regimes with different coefficients and variances, but the random components have the same autocorrelations. A test based on the normalized uniform (NU) residuals obtained from conditional probability integral transformations (CPIT) is given for testing for laghautocorrelation of residuals from full-rank normal errors regression models. The exact distribution theory of these values permits the evaluation of the significance level for this test. Some limited power comparisons are made with the Durbin-Watson (D-W) statistic for a one-sample simple linear regression model, and for this case the new NU test has power somewhat less than the exact D-W statistic, but much better than the D-W bounds test. Percentage points are given for tests for lags 1-6, and formulas are given that can treat all situations.

ISSN:0040-1706    

DOI:10.1080/00401706.1991.10484872

出版商:Taylor & Francis Group    

年代:1991

数据来源: Taylor

摘要:

Strategies are compared for development of a linear regression model and the subsequent assessment of its predictive ability. Simulations were performed as a designed experiment over a range of data structures. Approaches using a forward selection of variables resulted in slightly smaller prediction errors and less biased estimators of predictive accuracy than all possible subsets selection but often did not improve on the full model. Random and balanced data splitting resulted in increased prediction errors and estimators with large mean squared error. To maximize predictive accuracy while retaining a reliable estimate of that accuracy, it is recommended that the entire sample usually he used for model development and assessment.

ISSN:0040-1706    

DOI:10.1080/00401706.1991.10484873

出版商:Taylor & Francis Group    

年代:1991

数据来源: Taylor

ISSN:0040-1706    

DOI:10.1080/00401706.1991.10484874

出版商:Taylor & Francis Group    

年代:1991

数据来源: Taylor