-
作者：CHRISTENSEN, R; JOHNSON, W
作者单位：University of California System; University of California Davis
摘要：The relationship between survival times and covariates is modelled via the accelerated failure time model with a Dirichlet process prior assumed for the underlying baseline survival function. Emphasis is placed on estimating the survival distribution for future individuals with given covariates. A method of estimating regression coefficients which uses the marginal distribution for observed censored data is given. The proposed procedures are shown to perform reasonably on two artificial data s...
-
作者：MULLER, HG; STADTMULLER, U
作者单位：Ulm University
-
作者：WALDEN, AT
摘要：An important problem in the physical sciences is estimation of the innovations of a noninvertible linear process. Here the problem of deconvolving such a time series containing a phase-lead is investigated. A compensating phase-lag filter is sought, and is estimated by maximization of a chosen statistic or projection index. It is demonstrated by simulation that a statistic which is superior to another in the ideal case of a full-band time series can be substantially inferior in the band-limite...
-
作者：TONG, H
摘要：We have shown how a locally orthogonal reparameterization of an autoregressive model leads to the Levinson-Durbin algorithm.
-
作者：BANKS, DL
摘要：This paper describes a version of the Bayesian bootstrap that assigns random Dirichlet mass uniformly across statistically equivalent blocks. The method requires prior knowledge that the underlying distribution is continuous with known compact support. Under these conditions, the resulting analysis has three advantages over traditional bootstrap competitors. First, it takes account of the probability integral transformation. Secondly, it often enables exact expressions for the bootstrap distri...
-
作者：DAVIES, P; PHILLIPS, AJ
摘要：A modified Mann-Whitney statistic is used to construct a permuation test for comparing samples of paired data from two populations when the population identities of each pair are uncertain. An estimator of the probability of population misidentification within a pair is also proposed. This is based on the Lehmann family of distributions. Simulation results show the new methods are generally as more effective than existing normal theory unless there is strong correlation within pairs.
-
作者：SCHOTT, JR
摘要：The asymptotic variance of a statistic used to test the equality of the smallest latent roots of a correlation matrix is computed. This permits an approximation to its null distribution suggested by Lawley (1956). A simulation is used to compare the performance of this approximation with others currently in use.
-
作者：HALL, P; HANNAN, EJ
摘要：We use the concepts of stochastic complexity, description length, and model selection to develop data-based methods for choosing smoothing parameters in nonparametric density estimation. In the case of histogram estimators, we derive a simple, exact formula for stochastic complexity when the prior distribution of cell probabilities is uniform over the class of all possible choices. The formula depends only on the data and the smoothing parameter, which is readily chosen according to the criter...
-
作者：HALPERIN, M; LAN, KKG; HAMDY, MI
作者单位：George Washington University
摘要：An approach to multiple comparisons defines the error control objective as assuring with given probability that k or fewer out of N specific linear combinations of p means are significant when an overall null hypothesis is true, or that k or fewer confidence interval statements are incorrect. The case k=0 is basically the Scheffe (1953) method except for the consideration of a specified set of N linear combinations rather than all possible linear combinations. A theorectical result is obtained...
-
作者：SMITH, SP; HAMMOND, K
摘要：A robust analysis for field data is proposed using a stratified rank regression model. That analysis follows Pettitt (1982). Alternatively, a new technique is described where residuals are log gamma random variables. The log gamma distribution is approximately normal when the shape parameter is large and this parameter is taken as an integer which is either preselected or estimated by a simple grid search. Algorithms for estimating location parameters by maximizing the likelihood based on with...
