On Sample Reuse Methods for Spatial Data

García-Soidán, Pilar; Hall, Peter A.

doi:10.2307/2533113

Cited by 24 publications

(21 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In Section 3.1, we shall analyze discrete values time series data using the pairwise likelihood and J(θ) is suitably estimated by means of a reuse sampling procedure, called window subsampling (Carlstein 1986, Hall & Jing 1996, Garcia-Soidan & Hall 1997, Heagerty & Lele 1998, Lumley & Heagerty 1999. We briefly restore the basic procedure behind window subsampling strategies, in the case of time series data y = (y 1 , .…”

Section: Example (Continued)mentioning

confidence: 99%

A note on composite likelihood inference and model selection

Varin¹,

Vidoni²

2005

Biometrika

277

241

View full text Add to dashboard Cite

A composite likelihood consists in a combination of valid likelihood objects, usually related to small subsets of data. The merit of composite likelihood is to reduce the computational complexity so that it is possible to deal with large datasets and very complex models, even when the use of standard likelihood or Bayesian methods is not feasible. In this paper, we aim to suggest an integrated, general approach to inference and model selection using composite likelihood methods. In particular, we introduce an information criterion for model selection based on composite likelihood. Applications to modelling time series of counts through dynamic generalized linear models and to the analysis of the well-known Old Faithful geyser dataset are also given.

show abstract

Section: Example (Continued)mentioning

confidence: 99%

A note on composite likelihood inference and model selection

Varin¹,

Vidoni²

2005

Biometrika

277

241

View full text Add to dashboard Cite

show abstract

“…The sample variance of the replicates is used to estimate the asymptotic variance of the estimating function, which in turn is used to estimate the asymptotic variance of the target statistic. The same idea was also used by Sherman [25] and GarciaSoidan and Hall [15].…”

Section: Subsampling Estimation Of 1 +mentioning

confidence: 99%

On variance estimation in a negative binomial time series regression model

2012

Journal of Multivariate Analysis

View full text Add to dashboard Cite

a b s t r a c tWe study variance estimation in a negative binomial regression model for analyzing time series of counts, where serial dependence among the observed counts is introduced by an autocorrelated latent process. The regression coefficient vector is estimated by maximizing the pseudo-likelihood with the latent process suppressed. The resulting estimator is referred to as the generalized linear model estimator, and its consistency and asymptotic normality have been established by Davis and Wu [R.A. Davis, R. Wu, A negative binomial model for time series of counts, Biometrika 96 (2009) 735-749] when the latent process is stationary and strongly mixing. However, in order to perform valid statistical inferences about the regression coefficients, it is essential to develop a consistent estimation procedure for the asymptotic covariance matrix of the generalized linear model estimator. We propose two types of estimators using kernel-based and subsampling methods, and establish their consistency property. The results can be generalized straightforwardly to time series following a parameter-driven generalized linear model. Simulation study is conducted to evaluate the finite sample performance of the estimation methods.

show abstract

“…In this case the major challenge is the choice of the block length, which constitutes a trade-off in the bias-variance of estimates and is usually determined by the dependence structure of the data and the parameters under estimation (Davison and Hinkley, 1997;Hall et al, 1995). Garcia-Soidan and Hall (1997) proposed an alternative to the block bootstrap, by using a moving sampling window that translates to all possible locations within the data in order to estimate repeatedly the statistic of interest. Other more elaborated schemes within the block bootstrap framework are the use of random block sizes (Politis and Romano, 1994), and wrapping and overlapping blocks of data (Politis and Romano, 1993) in order to improve the bias and variance of the estimates.…”

Section: Resampling Schemesmentioning

confidence: 99%