Showing papers in "Computational Statistics & Data Analysis in 2000"
••
TL;DR: This paper proposes a new general approach, based on the methods of Hadi (1992a,1994) and Hadi and Simonoff (1993) that can be computed quickly — often requiring less than five evaluations of the model being fit to the data, regardless of the sample size.
506 citations
••
TL;DR: Computer-intensive non-parametric modelling procedures such as classification and regression trees (CART) and multivariate adaptive regression splines (MARS) can provide more informative and attractive models whose individual components can be displayed graphically.
142 citations
••
TL;DR: In this paper, a resistant rule, based on the linear combination ofquartiles, is proposed for identifying the existence of outliers in the Gaussian case, and the improvement occurs in terms of resistance and efficiency.
130 citations
••
TL;DR: A doubly linear adaptive fuzzy regression model is proposed, based on two linear models: a core regression model and a spread regression model, which “explains” the centers of the fuzzy observations, while the second one is for their spreads.
125 citations
••
TL;DR: An international database of case reports, each one describing a possible case of adverse drug reactions (ADRs), is maintained by the Uppsala Monitoring Centre (UMC), for the WHO international prognosis.
115 citations
••
TL;DR: In this paper, an EM algorithm for estimating the Poisson parameters of zero-inflated, zero-deflated, and standard Poisson models, when the zero observations are ignored, is presented.
111 citations
••
TL;DR: It is shown how the use of probability trees to store and to approximate probability potentials, and a careful selection of the deletion sequence, make this Monte-Carlo algorithm able to propagate over large networks with extreme probabilities.
105 citations
••
TL;DR: This work provides a framework to perform prediction in some types of binary random elds by using a Bayesian approach to map a binary outcome over a bounded region D of the plane and provides measures of prediction uncertainty amenable for binary outcomes.
95 citations
••
TL;DR: This article showed that adding small normal noise to replicate the success in the training set could slightly improve estimates in several common classification models, namely, nearest neighbor, neural networks, classification trees, and quadratic discriminant.
86 citations
••
TL;DR: In this paper, the performance of feed-forward back-propagation networks was evaluated under nine different experimental conditions, including the number of inputs and interactions, degree of censoring, proportional vs. non-proportional hazards, and sample size.
79 citations
••
TL;DR: The DEEPLOC algorithm as mentioned in this paper approximates the deepest location in higher dimensions by a multivariate generalization of the median, which can be seen as a kind of multivariate ranking.
••
TL;DR: A multivariate approach to binary segmentation in order to deal with more response variables and to explore dependency in multivariate data is provided.
••
TL;DR: In this paper, the sampling behavior of a nonparametric estimator of the overlapping coefficient was examined using Monte Carlo techniques and it was found that the bias of the estimator increases as the similarity of the distributions from which the samples are obtained increases.
••
TL;DR: In this article, the authors present an algorithm for computing the cumulative distribution function of the Kolmogorov-Smirnov test statistic D n in the all-parameters-known case.
••
TL;DR: The consequences for population inferences using popular methods for fitting nonlinear mixed effects models when the normality assumption is inappropriate and/or the model is misspecified are investigated.
••
TL;DR: In this article, the problem of blocking response surface designs when the block sizes are prepecified to control variation efficiently and the treatment set is chosen independently of the block structure is considered.
••
TL;DR: In this paper, the authors explore a class of quantile regression spline estimators of the quantiles regression function, which can be used to measure the effect of covariates not only in the center of a population, but also in the upper and lower tails of the population.
••
TL;DR: In this paper, classical and Bayesian methods for interval estimation of the ratio of two independent Poisson rates are examined and compared in terms of their exact coverage properties, and two methods to determine sampling effort requirements are derived.
••
TL;DR: In this article, two new adjustments for the R 2 -measure in Poisson regression models based on deviance residuals are presented and compared by simulation with population values, also applied to real data sets.
••
TL;DR: In this article, a data-driven procedure for obtaining parsimonious mixture model estimates or, conversely, kernel estimates with data driven local smoothing properties is described and investigated, where the main idea is to obtain a semiparametric estimate by alternating between the parametric and nonparametric viewpoints.
••
TL;DR: In this article, the authors consider three types of predictive regions for these predictors, i.e., the conditional percentile interval (CPI), the shortest conditional modal interval (SCMI), and the maximum conditional density region (MCDR).
••
TL;DR: In this paper, a maximum likelihood (ML)-based method and a two-stage bootstrap approach were proposed to construct confidence intervals for the ratio in the means of two independent populations which contain both log-normal and zero observations.
••
TL;DR: In this paper, the authors developed a framework to perform maximum likelihood estimation in the multinomial probit model using a Monte Carlo EM algorithm, which does not involve direct evaluation and maximization of the observed data likelihood.
••
TL;DR: The use of these algorithms in estimating the parameters of nonlinear regression models with several criteria like residual sum of squares, sum of absolute deviations and sum of trimmed squares are chosen.
••
TL;DR: Scheiner et al. as discussed by the authors used two resampling techniques, namely, the Jackknife and Bootstrap along with the Taylor series approximation and transformation method, for the construction of condence intervals.
••
TL;DR: In this paper, the diversity of estimates in regression analysis of some real data sets is discussed and a proposal for how to select from the diverse estimates of model is given and illustrated in examples of real data set.
••
TL;DR: In this article, the median-unbiased estimator was extended to the higher-order autoregressive processes, the nonnormal error term and inclusion of any exogenous variables.
••
TL;DR: The authors used Monte Carlo simulation to estimate the statistical power and Type I error rates of five procedures for testing the significance of a common risk difference in a set of independent 2×2 tables.
••
TL;DR: In this article, the authors compare the performance of bootstrap methods with other methods in terms of average coverage probability by Monte Carlo simulation and discuss the advantages of the bootstrap method.
••
TL;DR: In this article, the authors proposed a scale trend test based on Bartholomew's likelihood ratio test and a location-scale trend test using the sum of a location and a scale test statistic.