Professor Matt Wand is a Distinguished Professor of Statistics at University of Technology Sydney.
His latest research outputs are available on the website: http://matt-wand.utsacademics.info
He has held faculty appointments at Harvard University, Rice University, Texas A&M University, University of New South Wales and University of Wollongong. In 2008 Professor Wand became an elected Fellow of the Australian Academy of Science. He also has been awarded two Australian Academy of Science honorific awards for statistical research: the Moran Medal in 1997 for outstanding research by scientists under the age of 40 and the Hannan Medal in 2013 for career research in statistical science. In 2013 he was awarded the University of Technology Sydney, Chancellor's Medal for Exceptional Research. He received the 2013 Pitman Medal from the Statistical Society of Australia in recognition of outstanding achievement in, and contribution to, the discipline of Statistics. Professor Wand is an elected fellow of the American Statistical Association, the Institute of Mathematical Statistics and the Australian Mathematical Society.
Professor Wand has co-authored three books and more than 120 papers in statistics journals. He has several packages in the R language on the Comprehensive R Archive Network.
In 2002 Professor Wand was ranked 23 among highly cited authors in mathematics and statistics for the period 1991–2001. He has been a member of the ‘ISI Highly Cited Researchers’ lists. Since 2000 Professor Wand has been principal investigator on eight major grants. A recent one, an Australian Research Council Discovery Project, is titled ‘Fast Approximate Inference Methods: New Algorithms, Applications and Theory’ and will run for the years 2018–2020. Another is the `Australian Research Council Centre of Excellence for Mathematical and Statistical Frontiers' and is running during 2014-2021.
For more information visit his personal website http://matt-wand.utsacademics.info.
Matt serves as an associate editor for the Statistics journal: Australian and New Zealand Journal of Statistics.
He has previously served as an associate editor for the Journal of the American Statistical Association, Biometrika and Statistica Sinica.
He also participates in committee work within the Australian Academy of Science and the Australian Research Council.
Can supervise: YES
Professor Wand is chiefly interested in the development of statistical methodology for finding useful structure in large multivariate data sets.
Currently, Matt’s specific interests include: expectation propagation, message passing algorithms, variational approximate methods, statistical methods for streaming data, generalised linear mixed models and semiparametric regression.
He is also very interested in Statistical Computing and contributes to the field's main software repository — the ‘Comprehensive R Archive Network’.
37458 Advanced Bayesian Methods
This easy-to-follow applied book expands upon the authors' prior work on semiparametric regression to include the use of R software. In 2003, authors Ruppert and Wand co-wrote Semiparametric Regression with R.J. Carroll, which introduced the techniques and benefits of semiparametric regression in a concise and user-friendly fashion. Fifteen years later, semiparametric regression is applied widely, powerful new methodology is continually being developed, and advances in the R computing environment make it easier than ever before to carry out analyses.
Semiparametric Regression with R introduces the basic concepts of semiparametric regression with a focus on applications and R software. This volume features case studies from environmental, economic, financial, and other fields. The examples and corresponding code can be used or adapted to apply semiparametric regression to a wide range of problems. It contains more than fifty exercises, and the accompanying HRW package contains all datasets and scripts used in the book, as well as some useful R functions.
This book is suitable as a textbook for advanced undergraduates and graduate students, as well as a guide for statistically-oriented practitioners, and could be used in conjunction with Semiparametric Regression. Readers are assumed to have a basic knowledge of R and some exposure to linear models. For the underpinning principles, calculus-based probability, statistics, and linear algebra are desirable.
Ruppert, D, Wand, M & Carroll, RJ 2003, Semiparametric Regression, 1, Cambridge University Press, New York.
Wand, M & Jones, MC 1995, Kernel Smoothing, First, Chapman and Hall, London.
© 2020 Statistical Modeling Society. A two-level group-specific curve model is such that the mean response of each member of a group is a separate smooth function of a predictor of interest. The three-level extension is such that one grouping variable is nested within another one, and higher level extensions are analogous. Streamlined variational inference for higher level group-specific curve models is a challenging problem. We confront it by systematically working through two-level and then three-level cases and making use of the higher level sparse matrix infrastructure laid down in (Nolan and Wand (2020), ANZIAM Journal, doi: 10.1017/S1446181120000061). A motivation is analysis of data from ultrasound technology for which three-level group-specific curve models are appropriate. Whilst extension to the number of levels exceeding three is not covered explicitly, the pattern established by our systematic approach sheds light on what is required for even higher level group-specific curve models.
Kim, ASI & Wand, MP 2018, 'On expectation propagation for generalised, linear and mixed models', Australian and New Zealand Journal of Statistics, vol. 60, no. 1, pp. 75-102.View/Download from: Publisher's site
© 2018 Australian Statistical Publishing Association Inc. Published by John Wiley & Sons Australia Pty Ltd. Expectation propagation is a general approach to deterministic approximate Bayesian inference for graphical models, although its literature is confined mostly to machine learning applications. We investigate the utility of expectation propagation in generalised, linear, and mixed model settings. We show that, even though the algebra and computations are complicated, the notion of message passing on factor graphs affords streamlining of the required calculations and we list the algorithmic steps explicitly. Numerical studies indicate expectation propagation is marginally more accurate than a competing method for the models considered, but at the expense of bigger algebraic and computational overheads.
Liu, SH, Bobb, JF, Henn, BC, Schnaas, L, Tellez-Rojo, MM, Gennings, C, Arora, M, Wright, RO, Coull, BA & Wand, MP 2018, 'Modeling the health effects of time-varying complex environmental mixtures: Mean field variational Bayes for lagged kernel machine regression.', Environmetrics, vol. 29, no. 4.View/Download from: Publisher's site
There is substantial interest in assessing how exposure to environmental mixtures, such as chemical mixtures, affect child health. Researchers are also interested in identifying critical time windows of susceptibility to these complex mixtures. A recently developed method, called lagged kernel machine regression (LKMR), simultaneously accounts for these research questions by estimating effects of time-varying mixture exposures, and identifying their critical exposure windows. However, LKMR inference using Markov chain Monte Carlo methods (MCMC-LKMR) is computationally burdensome and time intensive for large datasets, limiting its applicability. Therefore, we develop a mean field variational Bayesian inference procedure for lagged kernel machine regression (MFVB-LKMR). The procedure achieves computational efficiency and reasonable accuracy as compared with the corresponding MCMC estimation method. Updating parameters using MFVB may only take minutes, while the equivalent MCMC method may take many hours or several days. We apply MFVB-LKMR to PROGRESS, a prospective cohort study in Mexico. Results from a subset of PROGRESS using MFVB-LKMR provide evidence of significant positive association between second trimester cobalt levels and z-scored birthweight. This positive association is heightened by cesium exposure. MFVB-LKMR is a promising approach for computationally efficient analysis of environmental health datasets, to identify critical windows of exposure to complex mixtures.
Pham, TH & Wand, MP 2018, 'Generalised additive mixed models analysis via gammSlice', Australian and New Zealand Journal of Statistics, vol. 60, no. 3, pp. 279-300.View/Download from: Publisher's site
© 2018 Australian Statistical Publishing Association Inc. Published by John Wiley & Sons Australia Pty Ltd. We demonstrate the use of our R package, gammSlice, for Bayesian fitting and inference in generalised additive mixed model analysis. This class of models includes generalised linear mixed models and generalised additive models as special cases. Accurate Bayesian inference is achievable via sufficiently large Markov chain Monte Carlo (MCMC) samples. Slice sampling is a key component of the MCMC scheme. Comparisons with existing generalised additive mixed model software shows that gammSlice offers improved inferential accuracy, albeit at the cost of longer computational time.
Rust, LT, Nizio, KD, Wand, MP & Forbes, SL 2018, 'Investigating the detection limits of scent-detection dogs to residual blood odour on clothing', Forensic Chemistry, vol. 9, pp. 62-75.View/Download from: Publisher's site
© 2018 Elsevier B.V. Blood-detection dogs are trained to locate blood evidence and search for potential crime scenes in cases where a cadaver may not be present. The locations of crime scenes are often ambiguous and evidence may not always be obvious during initial processing. In cases of foul play, a criminal may attempt to clean biological evidence from a crime scene; however, trace evidence that appears invisible to the naked eye may still be detectable. For example, it has been reported anecdotally that blood-detection dogs are capable of detecting blood on clothing that has been washed up to five times, or on surfaces which have been scrubbed clean. This study aimed to investigate the baseline detection limits of blood-detection dogs and cadaver-detection dogs to latent blood evidence on washed clothing and to compare the dogs' responses to current presumptive chemical and analytical techniques. Blood was deposited onto cotton swatches and washed up to five times with a standard household washing machine. Following washing, the cotton swatches were allowed to dry and presented to blood-detection and cadaver-detection dogs during law enforcement training. Replicates of these samples were tested with luminol spray and analysed using headspace solid phase microextraction – comprehensive two-dimensional gas chromatography – time-of-flight mass spectrometry (HS-SPME-GC×GC-TOFMS). Results indicated that the olfactory system of blood-detection and cadaver-detection dogs is a viable complementary technique to presumptive chemical tests and more sensitive than current scientific instrumentation, with some of the dogs able to detect blood after five washes but HS-SPME-GC×GC-TOFMS only able to detect blood after two washes or less. This limit of detection could likely be lowered for the dogs with further and more consistent training. Luminol was similarly able to detect blood washed up to five times, which indicates that the scenting abilities of these dogs can provide...
Wand, MP 2017, 'Fast Approximate Inference for Arbitrarily Large Semiparametric Regression Models via Message Passing', Journal of the American Statistical Association, vol. 112, no. 517, pp. 137-168.View/Download from: Publisher's site
© 2017 American Statistical Association. We show how the notion of message passing can be used to streamline the algebra and computer coding for fast approximate inference in large Bayesian semiparametric regression models. In particular, this approach is amenable to handling arbitrarily large models of particular types once a set of primitive operations is established. The approach is founded upon a message passing formulation of mean field variational Bayes that utilizes factor graph representations of statistical models. The underlying principles apply to general Bayesian hierarchical models although we focus on semiparametric regression. The notion of factor graph fragments is introduced and is shown to facilitate compartmentalization of the required algebra and coding. The resultant algorithms have ready-to-implement closed form expressions and allow a broad class of arbitrarily large semiparametric regression models to be handled. Ongoing software projects such as Infer.NET and Stan support variational-type inference for particular model classes. This article is not concerned with software packages per se and focuses on the underlying tenets of scalable variational inference algorithms. Supplementary materials for this article are available online.
Kim, SI & Wand, MP 2016, 'The explicit form of expectation propagation for a simple statistical model', Electronic Journal of Statistics, vol. 10, no. 1, pp. 550-581.View/Download from: Publisher's site
© 2016, Institute of Mathematical Statistics. All rights reserved. We derive the explicit form of expectation propagation for approximate deterministic Bayesian inference in a simple statistical model. The model corresponds to a random sample from the Normal distribution. The explicit forms, and their derivation, allow a deeper understanding of the issues and challenges involved in practical implementation of expectation propagation for statistical analyses. No auxiliary approximations are used: we follow the expectation propagation prescription exactly. A simulation study shows expectation propagation to be more accurate than mean field variational Bayes for larger sample sizes, but at the cost of considerably more algebraic and computational effort.
Lee, CY & Wand, MP 2016, 'Streamlined mean field variational Bayes for longitudinal and multilevel data analysis.', Biometrical journal. Biometrische Zeitschrift, vol. 58, pp. 868-895.View/Download from: Publisher's site
Streamlined mean field variational Bayes algorithms for efficient fitting and inference in large models for longitudinal and multilevel data analysis are obtained. The number of operations is linear in the number of groups at each level, which represents a two orders of magnitude improvement over the naïve approach. Storage requirements are also lessened considerably. We treat models for the Gaussian and binary response situations. Our algorithms allow the fastest ever approximate Bayesian analyses of arbitrarily large longitudinal and multilevel datasets, with little degradation in accuracy compared with Markov chain Monte Carlo. The modularity of mean field variational Bayes allows relatively simple extension to more complicated scenarios.
Lee, CY & Wand, MP 2016, 'Variational methods for fitting complex Bayesian mixed effects models to health data.', Statistics in medicine, vol. 35, no. 2, pp. 165-188.View/Download from: Publisher's site
We consider approximate inference methods for Bayesian inference to longitudinal and multilevel data within the context of health science studies. The complexity of these grouped data often necessitates the use of sophisticated statistical models. However, the large size of these data can pose significant challenges for model fitting in terms of computational speed and memory storage. Our methodology is motivated by a study that examines trends in cesarean section rates in the largest state of Australia, New South Wales, between 1994 and 2010. We propose a group-specific curve model that encapsulates the complex nonlinear features of the overall and hospital-specific trends in cesarean section rates while taking into account hospital variability over time. We use penalized spline-based smooth functions that represent trends and implement a fully mean field variational Bayes approach to model fitting. Our mean field variational Bayes algorithms allow a fast (up to the order of thousands) and streamlined analytical approximate inference for complex mixed effects models, with minor degradation in accuracy compared with the standard Markov chain Monte Carlo methods. Copyright © 2015 John Wiley & Sons, Ltd.
Rohde, D & Wand, MP 2016, 'Semiparametric mean field variational bayes: General principles and numerical issues', Journal of Machine Learning Research, vol. 17, no. 172, pp. 1-47.
© 2016 David Rohde and Matt P. Wand.We introduce the term semiparametric mean field variational Bayes to describe the relaxation of mean field variational Bayes in which some density functions in the product density restriction are pre-specified to be members of convenient parametric families. This notion has appeared in various guises in the mean field variational Bayes literature during its history and we endeavor to unify this important topic. We lay down a general framework and explain how previous relevant methodologies fall within this framework. A major contribution is elucidation of numerical issues that impact semiparametric mean field variational Bayes in practice.
© 2016 The Royal Statistical Society Making effective public policy decisions is challenging at the best of times, but especially in the context of environmental regulation, which typically requires managing opposing interests and strong opinions from industry and private citizens. In this case study, Louise Ryan, Matt Wand and Alan Malecki show how statistical analysis can help resolve conflict and inform effective decision-making under uncertainty.
Kayal, M, Vercelloni, J, Wand, MP & Adjeroud, M 2015, 'Searching for the best bet in life-strategy: A quantitative approach to individual performance and population dynamics in reef-building corals', ECOLOGICAL COMPLEXITY, vol. 23, pp. 73-84.View/Download from: Publisher's site
Menictas, M & Wand, MP 2015, 'VARIATIONAL INFERENCE FOR HETEROSCEDASTIC SEMIPARAMETRIC REGRESSION', AUSTRALIAN & NEW ZEALAND JOURNAL OF STATISTICS, vol. 57, no. 1, pp. 119-138.View/Download from: Publisher's site
We develop algorithms for performing semiparametric regression analysis in real time, with data processed as it is collected and made immediately available via modern telecommunications technologies. Our definition of semiparametric regression is quite broad and includes, as special cases, generalized linear mixed models, generalized additive models, geostatistical models, wavelet nonparametric regression models and their various combinations. Fast updating of regression fits is achieved by couching semiparametric regression into a Bayesian hierarchical model or, equivalently, graphical model framework and employing online mean field variational ideas. An Internet site attached to this article, realtime-semiparametric-regression.net, illustrates the methodology for continually arriving stock market, real estate, and airline data. Flexible real-time analyses based on increasingly ubiquitous streaming data sources stand to benefit. This article has online supplementary material.
Neville, SE, Ormerod, JT & Wand, MP 2014, 'Mean field variational Bayes for continuous sparse signal shrinkage: Pitfalls and remedies', Electronic Journal of Statistics, vol. 8, no. 1, pp. 1113-1151.View/Download from: Publisher's site
We investigate mean field variational approximate Bayesian inference for models that use continuous distributions, Horseshoe, Negative-Exponential-Gamma and Generalized Double Pareto, for sparse signal shrinkage. Our principal finding is that the most natural, and simplest, mean field variational Bayes algorithm can perform quite poorly due to posterior dependence among auxiliary variables. More sophisticated algorithms, based on special functions, are shown to be superior. Continued fraction approximations via Lentz's Algorithm are developed to make the algorithms practical.
Wand, MP 2014, 'Fully simplified multivariate normal updates in non-conjugate variational message passing', Journal of Machine Learning Research, vol. 15, pp. 1351-1369.
A family of prior distributions for covariance matrices is studied. Members of the family possess the attractive property of all standard deviation and correlation parameters being marginally noninformative for particular hyper-parameter choices. Moreove
We derive a variational inference procedure for approximate Bayesian inference in marginal longitudinal semiparametric regression. Fitting and inference is much faster than existing Markov chain Monte Carlo approaches. Numerical studies indicate that the new methodology is very accurate for the class of models under consideration. Copyright 2013 John Wiley & Sons Ltd
Wand, M, Ormerod, JT & Pham, T 2013, 'Mean field variational Bayesian inference for nonparametric regression with measurement error', Computational Statistics and Data Analysis, vol. 68, no. 1, pp. 375-387.View/Download from: Publisher's site
A fast mean field variational Bayes (MFVB) approach to nonparametric regression when the predictors are subject to classical measurement error is investigated. It is shown that the use of such technology to the measurement error setting achieves reasonable accuracy. In tandem with the methodological development, a customized Markov chain Monte Carlo method is developed to facilitate the evaluation of accuracy of the MFVB method.
Gloag, ES, Turnbull, L, Huang, A, Vallotton, P, Wang, H, Nolan, LM, Mililli, L, Hunt, C, Lu, J, Osvath, SR, Monahan, LG, Cavaliere, R, Charles, IG, Wand, M, Gee, M, Ranganathan, P & Whitchurch, CB 2013, 'Self-organization of bacterial biofilms is facilitated by extracellular DNA', Proceedings of the National Academy of Sciences of the United States of America, vol. 110, no. 28, pp. 11541-11546.View/Download from: Publisher's site
Twitching motility-mediated biofilm expansion is a complex, multicellular behavior that enables the active colonization of surfaces by many species of bacteria. In this study we have explored the emergence of intricate network patterns of interconnected trails that form in actively expanding biofilms of Pseudomonas aeruginosa. We have used high-resolution, phase-contrast time-lapse microscopy and developed sophisticated computer vision algorithms to track and analyze individual cell movements during expansion of P. aeruginosa biofilms. We have also used atomic force microscopy to examine the topography of the substrate underneath the expanding biofilm. Our analyses reveal that at the leading edge of the biofilm, highly coherent groups of bacteria migrate across the surface of the semisolid media and in doing so create furrows along which following cells preferentially migrate. This leads to the emergence of a network of trails that guide mass transit toward the leading edges of the biofilm. We have also determined that extracellular DNA (eDNA) facilitates efficient traffic flow throughout the furrow network by maintaining coherent cell alignments, thereby avoiding traffic jams and ensuring an efficient supply of cells to the migrating front. Our analyses reveal that eDNA also coordinates the movements of cells in the leading edge vanguard rafts and is required for the assembly of cells into the bulldozer aggregates that forge the interconnecting furrows. Our observations have revealed that large-scale self-organization of cells in actively expanding biofilms of P. aeruginosa occurs through construction of an intricate network of furrows that is facilitated by eDNA
Ormerod, JT & Wand, M 2012, 'Gaussian Variational Approximate Inference For Generalized Linear Mixed Models', Journal of Computational and Graphical Statistics, vol. 21, no. 1, pp. 2-17.View/Download from: Publisher's site
Variational approximation methods have become a mainstay of contemporary machine learning methodology, but currently have little presence in statistics. We devise an effective variational approximation strategy for fitting generalized linear mixed models
The agéd number theoretic concept of continued fractions can enhance certain Bayesian computations. The crux of this claim is due to continued fraction representations of numerically challenging special function ratios that arise in Bayesian computing. Continued fraction approximation via Lentz's Algorithm often leads to efficient and stable computation of such quantities.
Chacon, JE, Duong, T & Wand, M 2011, 'Asymptotics for general multivariate kernel density derivative estimators', Statistica Sinica, vol. 21, pp. 807-840.
We investigate kernel estimators of multivariate density derivative functions using general (or unconstrained) bandwidth matrix selectors. These density derivative estimators have been relatively less well researched than their density estimator analogues. A major obstacle for progress has been the intractability of the matrix analysis when treating higher order multivariate derivatives. With an alternative vectorization of these higher order derivatives, mathematical intractabilities are surmounted in an elegant and unified framework. The finite sample and asymptotic analysis of squared errors for density estimators are generalized to density derivative estimators. Moreover, we are able to exhibit a closed form expression for a normal scale bandwidth matrix for density derivative estimators. These normal scale bandwidths are employed in a numerical study to demonstrate the gain in performance of unconstrained selectors over their constrained counterparts.
Faes, C, Ormerod, JT & Wand, M 2011, 'Variational Bayesian inference for parametric and nonparametric regression with missing data', Journal of the American Statistical Association, vol. 105, no. 495, pp. 959-971.View/Download from: Publisher's site
Bayesian hierarchical models are attractive structures for conducting regression analyses when the data are subject to missingness. However, the requisite probability calculus is challenging and Monte Carlo methods typically are employed. We develop an alternative approach based on deterministic variational Bayes approximations. Both parametric and nonparametric regression are considered. Attention is restricted to the more challenging case of missing predictor data. We demonstrate that variational Bayes can achieve good accuracy, but with considerably less computational overhead. The main ramification is fast approximate Bayesian inference in parametric and nonparametric regression models with missing data.
We introduce variational Bayes methods for fast approximate inference in functional regression analysis. Both the standard cross-sectional and the increasingly common longitudinal settings are treated. The methodology allows Bayesian functional regression analyses to be conducted without the computational overhead of Monte Carlo methods. Confidence intervals of the model parameters are obtained both using the approximate variational approach and nonparametric resampling of clusters. The latter approach is possible because our variational Bayes functional regression approach is computationally efficient. A simulation study indicates that variational Bayes is highly accurate in estimating the parameters of interest and in approximating the Markov chain Monte Carlo-sampled joint posterior distribution of the model parameters. The methods apply generally, but are motivated by a longitudinal neuroimaging study of multiple sclerosis patients. Code used in simulations is made available as a web-supplement.
Hall, P, Ormerod, JT & Wand, M 2011, 'Theory of Gaussian variational approximation for a Poisson mixed model', Statistica Sinica, vol. 21, no. 1, Special Issue, pp. 369-389.
Likelihood-based inference for the parameters of generalized linear mixed models is hindered by the presence of intractable integrals. Gaussian variational approximation provides a fast and effective means of approximate inference. We provide some theory for this type of approximation for a simple Poisson mixed model. In particular, we establish consistency at rate m(-1/2) + n(-1), where in is the number of groups and n is the number of repeated measurements.
Hall, P, Pham, T, Wand, MP & Wang, SSJ 2011, 'Asymptotic normality and valid inference for Gaussian variational approximation', Annals of Statistics, vol. 39, no. 5, pp. 2502-2532.View/Download from: Publisher's site
We derive the precise asymptotic distributional behavior of Gaussian
variational approximate estimators of the parameters in a single-predictor
Poisson mixed model. These results are the deepest yet obtained concerning the
statistical properties of a variational approximation method. Moreover, they
give rise to asymptotically valid statistical inference. A simulation study
demonstrates that Gaussian variational approximate confidence intervals possess
good to excellent coverage properties, and have a similar precision to their
exact likelihood counterparts.
Neville, S, Palmer, M & Wand, M 2011, 'Generalized Extreme Value Additive Model Analysis Via Mean Field Variational Bayes', Australian & New Zealand Journal of Statistics, vol. 53, no. 3, pp. 305-330.View/Download from: Publisher's site
We develop Mean Field Variational Bayes methodology for fast approximate inference in Bayesian Generalized Extreme Value additive model analysis. Such models are useful for flexibly assessing the impact of continuous predictor variables on sample extreme
Wand, M & Ormerod, JT 2011, 'Penalized wavelets: Embedding wavelets into semiparametric regression', Electronic Journal of Statistics, vol. 5, no. 1, pp. 1654-1717.View/Download from: Publisher's site
We introduce the concept of penalized wavelets to facilitate seamless embedding of wavelets into semiparametric regression models. In particular, we show that penalized wavelets are analogous to penalized splines; the latter being the established approach to function estimation in semiparametric regression. They differ only in the type of penalization that is appropriate. This fact is not borne out by the existing wavelet literature, where the regression modelling and fitting issues are overshadowed by computational issues such as efficiency gains afforded by the Discrete Wavelet Transform and partially obscured by a tendency to work in the wavelet coefficient space. With penalized wavelet structure in place, we then show that fitting and inference can be achieved via the same general approaches used for penalized splines: penalized least squares, maximum likelihood and best prediction within a frequentist mixed model framework, and Markov chain Monte Carlo and mean field variational Bayes within a Bayesian framework. Penalized wavelets are also shown have a close relationship with wide data (p»n) regression and benefit from ongoing research on that topic
We develop strategies for mean eld variational Bayes approximate inference for Bayesian hierarchical models containing elaborate distributions. We loosely dene elaborate distributions to be those having more complicated forms compared with common distributions such as those in the Normal and Gamma families. Examples are Asymmetric Laplace, Skew Normal and Generalized Ex- treme Value distributions. Such models suer from the diculty that the param- eter updates do not admit closed form solutions. We circumvent this problem through a combination of (a) specially tailored auxiliary variables, (b) univariate quadrature schemes and (c) nite mixture approximations of troublesome den- sity functions. An accuracy assessment is conducted and the new methodology is illustrated in an application
We demonstrate and critique the new Bayesian inference package Infer.NET in terms of its capacity for statistical analyses. Infer.NET differs from the well-known BUGS Bayesian inference packages in that its main engine is the variational Bayes family of deterministic approximation algorithms rather than Markov chain Monte Carlo. The underlying rationale is that such deterministic algorithms can handle bigger problems due to their increased speed, despite some loss of accuracy. We find that Infer.NET is a well-designed computational framework and offers significant speed advantages over BUGS. Nevertheless, the current release is limited in terms of the breadth of models it can handle, and its inference is sometimes inaccurate. Supplemental materials accompany the online version of this article.
Al Kadiri, M, Carroll, RJ & Wand, M 2010, 'Marginal longitudinal semiparametric regression via penalized splines', Statistics & Probability Letters, vol. 80, no. 15-16, pp. 1242-1252.View/Download from: Publisher's site
We study the marginal longitudinal nonparametric regression problem and some of its semiparametric extensions. We point out that, while several elaborate proposals for efficient estimation have been proposed, a relative simple and straightforward one, based on penalized splines, has not. After describing our approach, we then explain how Gibbs sampling and the BUGS software can be used to achieve quick and effective implementation. Illustrations are provided for nonparametric regression and additive models.
We devise a classification algorithm based on generalized linear mixed model (GLMM) technology. The algorithm incorporates spline smoothing, additive model-type structures and model selection. For reasons of speed we employ the Laplace approximation, rather than Monte Carlo methods. Tests on real and simulated data show the algorithm to have good classification performance. Moreover, the resulting classifiers are generally interpretable and parsimonious.
Marley, JK & Wand, M 2010, 'Non-standard semiparametric regression via BRugs', Journal of Statistical Software, vol. 37, no. 5, pp. 1-30.
We provide several illustrations of Bayesian semiparametric regression analyses in the BRugs package. BRugs facilitates use of the BUGS inference engine from the R computing environment and allows analyses to be managed using scripts. The examples are chosen to represent an array of non-standard situations, for which mixed model software is not viable. The situations include: the response variable being outside of the one-parameter exponential family, data subject to missingness, data subject to measurement error and parameters entering the model via an index.
Background High-throughput flow cytometry experiments produce hundreds of large multivariate samples of cellular characteristics. These samples require specialized processing to obtain clinically meaningful measurements. A major component of this processing is a form of cell subsetting known as gating. Manual gating is time-consuming and subjective. Good automatic and semi-automatic gating algorithms are very beneficial to high-throughput flow cytometry.
Variational approximations facilitate approximate inference for the parameters in complex statistical models and provide fast, deterministic alternatives to Monte Carlo methods. However, much of the contemporary literature on variational approximations is in Computer Science rather than Statistics, and uses terminology, notation, and examples from the former field. In this article we explain variational approximation in statistical terms. In particular, we illustrate the ideas of variational approximation using examples that are familiar to statisticians.
Samworth, RJ & Wand, M 2010, 'Asymptotics and optimal bandwidth selection for highest density region estimation', Annals of Statistics, vol. 38, no. 3, pp. 1767-1792.View/Download from: Publisher's site
We study kernel estimation of highest-density regions (HDR). Our main contributions are two-fold. First, we derive a uniform-in-bandwidth asymptotic approximation to a risk that is appropriate for HDR estimation. This approximation is then used to derive a bandwidth selection rule for HDR estimation possessing attractive asymptotic properties. We also present the results of numerical studies that illustrate the benefits of our theory and methodology.
Wand, MP & Ormerod, JT 2010, 'ON SEMIPARAMETRIC REGRESSION WITH O'SULLIVAN PENALISED SPLINES (vol 50, pg 179, 2008)', AUSTRALIAN & NEW ZEALAND JOURNAL OF STATISTICS, vol. 52, no. 2, pp. 239-239.View/Download from: Publisher's site
Duong, T, Koch, I & Wand, M 2009, 'Highest density difference region estimation with application to flow cytometric data', Biometrical Journal, vol. 51, no. 3, pp. 504-521.View/Download from: Publisher's site
Motivated by the needs of scientists using flow cytometry, we study the problem of estimating the region where two multivariate samples differ in density. We call this problem highest density difference region estimation and recognise it as a two-sample analogue of highest density region or excess set estimation. Flow cytometry samples are typically in the order of 10 000 and 100 000 and with dimension ranging from about 3 to 20. The industry standard for the problem being studied is called Frequency Difference Gating, due to Roederer and Hardy (2001). After couching the problem in a formal statistical framework we devise an alternative estimator that draws upon recent statistical developments such as patient rule induction methods. Improved performance is illustrated in simulations. While motivated by flow cytometry, the methodology is suitable for general multivariate random samples where density difference regions are of interest
Gottardo, R, Brinkman, RR, Luta, G, Luta, G & Wand, MP 2009, 'Recent bioinformatics advances in the analysis of high throughput flow cytometry data.', Advances in bioinformatics, p. 461763.View/Download from: Publisher's site
High-content flow cytometric screening (FC-HCS) is a 21st Century technology that combines robotic fluid handling, flow cytometric instrumentation, and bioinformatics software, so that relatively large numbers of flow cytometric samples can be processed and analysed in a short period of time. We revisit a recent application of FC-HCS to the problem of cellular signature definition for acute graft-versus-host-disease. Our focus is on automation of the data processing steps using recent advances in statistical methodology. We demonstrate that effective results, on par with those obtained via manual processing, can be achieved using our automatic techniques. Such automation of FC-HCS has the potential to drastically improve diagnosis and biomarker identification.
Two areas of research longitudinal data analysis and kernel machines have large, but mostly distinct, literatures. This article shows explicitly that both fields have much in common with each other. In particular, many popular longitudinal data fitting procedures are special types of kernel machines. These connections have the potential to provide fruitful cross-fertilization between longitudinal data analytic and kernel machine methodology
Semiparametric regression is a fusion between parametric regression and nonparametric regression that integrates low-rank penalized splines, mixed model and hierarchical Bayesian methodology thus allowing more streamlined handling of longitudinal and spatial correlation. We review progress in the field over the five-year period between 2003 and 2007. We find semiparametric regression to be a vibrant field with substantial involvement and activity, continual enhancement and widespread application.
The t-distribution allows the incorporation of outlier robustness into statistical models while retaining the elegance of likelihood-based inference. In this paper, we develop and implement a linear mixed model for the general design of the linear mixed model using the univariate t-distribution. This general design allows a considerably richer class of models to be fit than is possible with existing methods. Included in this class are semi-parametric regression and smoothing and spatial models.
Semiparametric regression models that use spline basis functions with penalization have graphical model representations. This link is more powerful than previously established mixed model representations of semiparametric regression, as a larger class of models can be accommodated. Complications such as missingness and measurement error are more naturally handled within the graphical model architecture. Directed acyclic graphs, also known as Bayesian networks, play a prominent role. Graphical model-based Bayesian `inference engines, such as bugs and vibes, facilitate fitting and inference. Underlying these are Markov chain Monte Carlo schemes and recent developments in variational approximation theory and methodology
Duong, T, Cowling, A, Koch, I & Wand, M 2008, 'Feature significance for multivariate kernel density estimation', Computational Statistics and Data Analysis, vol. 52, no. 9, pp. 4225-4242.View/Download from: Publisher's site
Multivariate kernel density estimation provides information about structure in data. Feature significance is a technique for deciding whether featuressuch as local extremaare statistically significant. This paper proposes a framework for feature significance in d-dimensional data which combines kernel density derivative estimators and hypothesis tests for modal regions. For the gradient and curvature estimators distributional properties are given, and pointwise test statistics are derived. The hypothesis tests extend the two-dimensional feature significance ideas of Godtliebsen et al. [Godtliebsen, F., Marron, J.S., Chaudhuri, P., 2002. Significance in scale space for bivariate density estimation. Journal of Computational and Graphical Statistics 11, 121]. The theoretical framework is complemented by novel visualization for three-dimensional data. Applications to real data sets show that tests based on the kernel curvature estimators perform well in identifying modal regions. These results can be enhanced by corresponding tests with kernel gradient estimators.
Fan, Y, Leslie, DS & Wand, M 2008, 'Generalised linear mixed model analysis via sequential Monte Carlo sampling', Electronic Journal of Statistics, vol. 2, pp. 916-938.View/Download from: Publisher's site
We present a sequential Monte Carlo sampler algorithm for the Bayesian analysis of generalised linear mixed models (GLMMs). These models support a variety of interesting regression-type analyses, but performing inference is often extremely difficult, even when using the Bayesian approach combined with Markov chain Monte Carlo (MCMC). The Sequential Monte Carlo sampler (SMC) is a new and general method for producing samples from posterior distributions. In this article we demonstrate use of the SMC method for performing inference for GLMMs. We demonstrate the effectiveness of the method on both simulated and real data, and find that sequential Monte Carlo is a competitive alternative to the available MCMC techniques.
Kuo, FY, Dunsmuir, WT, Sloan, IH, Wand, M & Womersley, RS 2008, 'Quasi-Monte Carlo for highly structured generalised response models', Methodology and Computing in Applied Probability, vol. 10, no. 2, pp. 239-275.View/Download from: Publisher's site
Highly structured generalised response models, such as generalised linear mixed models and generalised linear models for time series regression, have become an indispensable vehicle for data analysis and inference in many areas of application. However, their use in practice is hindered by high-dimensional intractable integrals. Quasi-Monte Carlo (QMC) is a dynamic research area in the general problem of high-dimensional numerical integration, although its potential for statistical applications is yet to be fully explored. We survey recent research in QMC, particularly lattice rules, and report on its application to highly structured generalised response models. New challenges for QMC are identified and new methodologies are developed. QMC methods are seen to provide significant improvements compared with ordinary Monte Carlo methods.
We study computational issues for support vector classification with penalised spline kernels. We show that, compared with traditional kernels, computational times can be drastically reduced in large problems making such problems feasible for sample sizes as large as ~106. The optimisation technology known as interior point methods plays a central role. Penalised spline kernels are also shown to allow simple incorporation of low-dimensional structure such as additivity. This can aid both interpretability and performance.
We consider additive models fitting and inference when the response variable is a sample extreme. Non-linear covariate effects are handled using the mixed model representation of penalised splines. A fitting algorithm based on likelihood approximations is derived. The efficacy of the resulting methodology is demonstrated via application to simulated and real data.
Semiparametric mixed model analysis benefits from variability estimates such as standard errors of effect estimates and variability bars to accompany curve estimates. We show how the underlying variance calculations can be done extremely efficiently compared with the direct naïve approach. These streamlined calculations are linear in the number of subjects, representing a two orders of magnitude improvement.
Wand, M & Ormerod, JT 2008, 'On semiparametric regression with O'Sullivan penalised splines', Australian & New Zealand Journal of Statistics, vol. 50, no. 2, pp. 179-198.View/Download from: Publisher's site
An exposition on the use of O'Sullivan penalized splines in contemporary semiparametric regression, including mixed model and Bayesian formulations, is presented. O'Sullivan penalized splines are similar to P-splines, but have the advantage of being a direct generalization of smoothing splines. Exact expressions for the O'Sullivan penalty matrix are obtained. Comparisons between the two types of splines reveal that O'Sullivan penalized splines more closely mimic the natural boundary behaviour of smoothing splines. Implementation in modern computing environments such asMatlab,r andbugs is discussed.
This paper develops inference for the significance of features such as peaks and valleys observed in additive modeling through an extension of the SiZer-type methodology of Chaudhuri and Marron (1999) and Godtliebsen et al. (2002, 2004) to the case where the outcome is discrete. We consider the problem of determining the significance of features such as peaks or valleys in observed covariate effects both for the case of additive modeling where the main predictor of interest is univariate as well as the problem of studying the significance of features such as peaks, inclines, ridges and valleys when the main predictor of interest is geographical location. We work with low rank radial spline smoothers to allow to the handling of sparse designs and large sample sizes. Reducing the problem to a Generalised Linear Mixed Model (GLMM) framework enables derivation of simulation-based critical value approximations and guards against the problem of multiple inferences over a range of predictor values. Such a reduction also allows for easy adjustment for confounders including those which have an unknown or complex effect on the outcome. A simulation study indicates that our method has satisfactory power. Finally, we illustrate our methodology on several data sets.
Oakes, SR, Robertson, FG, Kench, JG, Gardiner-Garden, M, Wand, M, Freen, JE & Ormandy, CJ 2007, 'Loss of mammary epithelial prolactin receptor delays tumor formation by reducing cell proliferation in low-grade preinvasive lesions', Oncogene, vol. 26, no. 4, pp. 543-553.View/Download from: Publisher's site
Top quartile serum prolactin levels confer a twofold increase in the relative risk of developing breast cancer. Prolactin exerts this effect at an ill defined point in the carcinogenic process, via mechanisms involving direct action via prolactin receptors within mammary epithelium and/or indirect action through regulation of other hormones such as estrogen and progesterone. We have addressed these questions by examining mammary carcinogenesis in transplants of mouse mammary epithelium expressing the SV40T oncogene, with or without the prolactin receptor, using host animals with a normal endocrine system. In prolactin receptor knockout transplants the area of neoplasia was significantly smaller (7 versus 17%; P<0.001 at 22 weeks and 7 versus 14%; P=0.009 at 32 weeks). Low-grade neoplastic lesions displayed reduced BrdU incorporation rate (11.3 versus 17% P=0.003) but no change in apoptosis rate. Tumor latency increased (289 days versus 236 days, P<0.001). Tumor frequency, growth rate, morphology, cell proliferation and apoptosis were not altered. Thus, prolactin acts directly on the mammary epithelial cells to increase cell proliferation in preinvasive lesions, resulting in more neoplasia and acceleration of the transition to invasive carcinoma. Targeting of mammary prolactin signaling thus provides a strategy to prevent the early progression of neoplasia to invasive carcinoma.
The Fisher information for the canonical link exponential family generalised linear mixed model is derived. The contribution from the fixed effects parameters is shown to have a particularly simple form.
Werneck, GL, Costa, CHN, Walker, AM, David, JR, Wand, M & Maguire, JH 2007, 'Multilevel modelling of the incidence of visceral leishmaniasis in Teresina, Brazil', EPIDEMIOLOGY AND INFECTION, vol. 135, no. 2, pp. 195-201.View/Download from: Publisher's site
Asthma researchers have found some evidence that geographical variations in susceptibility to asthma could reflect the effect of community level factors such as exposure to violence. Our methodology was motivated by a study of age at onset of asthma among children of inner-city neighbourhoods in East Boston. Cox's proportional hazards model was not appropriate since there was not enough information about the nature of geographical variations so as to impose a parametric relationship. In addition, some of the known risk factors were believed to have non-linear log-hazard ratios. We extend the geoadditive models of Kamman and Wand to the case where the outcome measure is a possibly censored time to event. We reduce the problem to one of fitting a Poisson mixed model by using Poisson approximations in conjunction with a mixed model formulation of generalized additive modelling. Our method allows for low-rank additive modelling, provides likelihood-based estimation of all parameters including the amount of smoothing and can be implemented using standard software. We illustrate our method on the East Boston data.
Two data analytic research areaspenalized splines and reproducing kernel methodshave become very vibrant since the mid-1990s. This article shows how the former can be embedded in the latter via theory for reproducing kernel Hilbert spaces. This connection facilitates cross-fertilization between the two bodies of research. In particular, connections between support vector machines and penalized splines are established. These allow for significant reductions in computational complexity, and easier incorporation of special structure such as additivity.
Wand, M 2006, 'Support vector machine classification', Parabola, vol. 42, no. 2, pp. 21-37.
Werneck, GL, Costa, CH, Walker, AM, David, JR, Wand, M & Maguire, JH 2006, 'Multilevel modelling of the incidence of visceral leishmaniasis in Teresina, Brazil', Epidemiology and Infection, vol. 135, no. 2, pp. 195-201.View/Download from: Publisher's site
Epidemics of visceral leishmaniasis (VL) in major Brazilian cities are new phenomena since 1980. As determinants of transmission in urban settings probably operate at different geographic scales, and information is not available for each scale, a multilevel approach was used to examine the effect of canine infection and environmental and socio-economic factors on the spatial variability of incidence rates of VL in the city of Teresina. Details on an outbreak of greater than 1200 cases of VL in Teresina during 19931996 were available at two hierarchical levels: census tracts (socio-economic characteristics, incidence rates of human VL) and districts, which encompass census tracts (prevalence of canine infection). Remotely sensed data obtained by satellite generated environmental information at both levels. Data from census tracts and districts were analysed simultaneously by multilevel modelling. Poor socio-economic conditions and increased vegetation were associated with a high incidence of human VL. Increasing prevalence of canine infection also predicted a high incidence of human VL, as did high prevalence of canine infection before and during the epidemic. Poor socio-economic conditions had an amplifying effect on the association between canine infection and the incidence of human VL. Focusing interventions on areas with characteristics identified by multilevel analysis could be a cost-effective strategy for controlling VL. Because risk factors for infectious diseases operate simultaneously at several levels and ecological data usually are available at different geographical scales, multilevel modelling is a valuable tool for epidemiological investigation of disease transmission
Linear mixed models are able to handle an extraordinary range of complications in regression-type analyses. Their most common use is to account for within-subject correlation in longitudinal data analysis. They are also the standard vehicle for smoothing spatial count data. However, when treated in full generality, mixed models can also handle spline-type smoothing and closely approximate kriging. This allows for nonparametric regression models (e.g., additive models and varying coefficient models) to be handled within the mixed model framework. The key is to allow the random effects design matrix to have general structure; hence our label general design. For continuous response data, particularly when Gaussianity of the response is reasonably assumed, computation is now quite mature and supported by the R, SAS and S-PLUS packages.
Crainiceanu, CM, Ruppert, D & Wand, M 2005, 'Bayesian analysis for penalized spline regression using WinBUGS', Journal of Statistical Software, vol. 14, no. 14.
Penalized splines can be viewed as BLUPs in a mixed model framework, which allows the use of mixed model software for smoothing. Thus, software originally developed for Bayesian analysis of mixed models can be used for penalized spline regression. Bayesian inference for nonparametric models enjoys the flexibility of nonparametric models and the exact inference provided by the Bayesian inferential machinery. This paper provides a simple, yet comprehensive, set of programs for the implementation of nonparametric Bayesian analysis in WinBUGS. Good mixing properties of the MCMC chains are obtained by using low-rank thin-plate splines, while simulation times per iteration are reduced employing WinBUGS specific computational tricks.
Penalised-spline-based additive models allow a simple mixed model representation where the variance components control departures from linear models. The smoothing parameter is the ratio of the random-coefficient and error variances and tests for linear regression reduce to tests for zero random-coefficient variances. We propose exactlikelihood and restricted likelihood ratio tests for testing polynomial regression versus a general alternative modelled by penalised splines. Their spectral decompositions are used as the basis of fast simulation algorithms. We derive the asymptotic local power properties of the tests under weak conditions. In particular we characterise the local alternatives that are detected with asymptotic probability one. Confidence intervals for the smoothing parameter are obtained by inverting the tests for a fixed smoothing parameter versus a general alternative. We discuss F and R tests and show that ignoring the variability in the smoothing parameter estimator can have a dramatic effect on their null distributions. The powers of several known tests are investigated and a small set of tests with good power properties is identified. The restricted likelihood ratio test is among the best in terms of power
Durban, M, Harezlak, J, Wand, M & Carroll, RJ 2005, 'Simple fitting of subject-specific curves for longitudinal data', Statistics in Medicine, vol. 24, no. 8, pp. 1153-1167.View/Download from: Publisher's site
We present a simple semiparametric model for tting subject-specic curves for longitudinal data. Individual curves are modelled as penalized splines with random coecients. This model has a mixed model representation, and it is easily implemented in standard statistical software. We conduct an analysis of the long-term eect of radiation therapy on the height of children suering from acute lymphoblastic leukaemia using penalized splines in the framework of semiparametric mixed eects models. The analysis revealed signicant dierences between therapies and showed that the growth rate of girls in the study cannot be fully explained by the group-average curve and that individual curves are necessary to reect the individual response to treatment. We also show how to implement these models in S-PLUS and R in the appendi
Ganguli, B, Staudenmayer, J & Wand, M 2005, 'Additive models with predictors subject to measurement error', Australian & New Zealand Journal of Statistics, vol. 47, no. 2, pp. 193-202.View/Download from: Publisher's site
This paper develops a likelihood-based method for fitting additive models in the presence of measurement error. It formulates the additive model using the linear mixed model representation of penalized splines. In the presence of a structural measurement error model, the resulting likelihood involves intractable integrals, and a Monte Carlo expectation maximization strategy is developed for obtaining estimates. The method's performance is illustrated with a simulation study.
Salganik, MP, Hardie, DL, Swart, B, Dandie, GW, Zola, H, Shaw, S, Shapiro, H, Tinckam, K, Milford, EL & Wand, M 2005, 'Detecting antibodies with similar reactivity patterns in the HLDA8 blind panel of flow cytometry data', Journal Of Immunological Methods, vol. 305, no. 1, pp. 67-74.View/Download from: Publisher's site
The blind panel collected for the 8th Human Leucocyte Differentiation Antigens Workshop (HLDA8; http://www.hlda8.org) included 49 antibodies of known CD specificities and 76 antibodies of unknown specificity. We have identified groups of antibodies showing similar patterns of reactivity that need to be investigated by biochemical methods to evaluate whether the antibodies within these groups are reacting with the same molecule. Our approach to data analysis was based on the work of Salganik et al. (in press) [Salganik, M.P., Milford E.L., Hardie D.L., Shaw, S., Wand, M.P., in press. Classifying antibodies using flow cytometry data: class prediction and class discovery
Salganik, MP, Milford, EL, Hardie, DL, Shaw, S & Wand, M 2005, 'Classifying antibodies using flow cytometry data: class prediction and class discovery', Biometrical Journal, vol. 47, no. 5, pp. 740-754.View/Download from: Publisher's site
Classifying monoclonal antibodies, based on the similarity of their binding to the proteins (antigens) on the surface of blood cells, is essential for progress in immunology, hematology and clinical medicine. The collaborative efforts of researchers from many countries have led to the classification of thousands of antibodies into 247 clusters of differentiation (CD). Classification is based on flow cytometry and biochemical data. In preliminary classifications of antibodies based on flow cytometry data, the object requiring classification (an antibody) is described by a set of random samples from unknown densities of fluorescence intensity. An individual sample is collected in the experiment, where a population of cells of a certain type is stained by the identical fluorescently marked replicates of the antibody of interest. Samples are collected for multiple cell types. The classification problems of interest include identifying new CDs (class discovery or unsupervised learning) and assigning new antibodies to the known CD clusters (class prediction or supervised learning). These problems have attracted limited attention from statisticians. We recommend a novel approach to the classification process in which a computer algorithm suggests to the analyst the subset of the most appropriate classifications of an antibody in class prediction problems or the most similar pairs/groups of antibodies in class discovery problems. The suggested algorithm speeds up the analysis of a flow cytometry data by a factor 1020. This allows the analyst to focus on the interpretation of the automatically suggested preliminary classification solutions and on planning the subsequent biochemical experiments
Swart, B, Salganik, MP, Wand, M, Tinckam, K, Milford, EL, Drbal, K, Angelisova, P, Horejsi, V, Macardle, P, Bailey, S, Hollemweguer, E, Hodge, G, Nairn, J, Millard, D, Dagdeviren, A, Dandie, GW & Zola, H 2005, 'The HLDA8 blind panel: Findings and conclusions', Journal Of Immunological Methods, vol. 305, no. 1, pp. 75-83.View/Download from: Publisher's site
There were over 600 antibodies submitted to HLDA8, with many of unknown specificity. Of these, 101 antibodies were selected for a blind panel study that also included 5 negative controls and 27 positive controls of known CD specificity making a total of 133 antibodies in the final panel. Of the 101 unknowns, 31 antibodies were identified during the course of this blind panel study as being specific for known molecules and included some specific for MHC class II antigens, CD45 isoforms and the Dombrock antigen. Several antibody pairs among those in the blind panel were found to have very similar staining patterns and were therefore compared by immunohistochemical and/or Western blot analyses for identity.
Myatt, TA, Johnston, SJ, Zhengfa, Z, Wand, M, Kebadze, T, Rudnick, S & Milton, DK 2004, 'Detection of airborne rhinovirus and its relation to outdoor air supply in office environments', American Journal of Respiratory and Critical Care Medicine, vol. 169, pp. 1187-1190.View/Download from: Publisher's site
Ngo, L & Wand, M 2004, 'Smoothing with mixed model software', Journal of Statistical Software, vol. 9, no. 1.
Salganik, MP, Wand, M & Lange, N 2004, 'Comparison of feature significance quantile approximations', Australian & New Zealand Journal of Statistics, vol. 46, pp. 569-581.View/Download from: Publisher's site
Wright, R, Finn, P, Contreras, JP, Cohen, S, Wright, RO, Staudenmayer, J, Wand, M, Perkins, D, Weiss, S & Gold, DR 2004, 'Chronic caregiver stress and IgE expression, allergen-induced proliferation, and cytokine profiles in a birth cohort predisposed to atopy', Journal Of Allergy And Clinical Immunology, vol. 113, no. 6, pp. 1051-1057.View/Download from: Publisher's site
Contreras, JP, Ly, NR, Gold, DR, He, HZ, Wand, M, Weiss, ST, Perkins, DL, Platts-Mills, TAE & Finn, PW 2003, 'Allergen-induced cytokine production, atopic disease, IgE, and wheeze in children', JOURNAL OF ALLERGY AND CLINICAL IMMUNOLOGY, vol. 112, no. 6, pp. 1072-1077.View/Download from: Publisher's site
Hauser, R, Rice, TM, Krishha Murthy, GG, Wand, M, Lewis, D, Bledsoe, T & Paulauskis, J 2003, 'The upper airway response to pollen is enhanced by exposure to combustion particulates: A pilot human experimental challenge study', Environmental Health Perspectives, vol. 111, no. 5, pp. 676-680.View/Download from: Publisher's site
Hauser, R, Rice, TM, Murthy, GGK, Wand, MP, Lewis, D, Bledsoe, T & Paulauskis, J 2003, 'The upper airway response to pollen is enhanced by exposure to combustion particulates: A pilot human experimental challenge study', Environmental Health Perspectives, vol. 111, no. 4, pp. 472-477.View/Download from: Publisher's site
Although human experimental studies have shown that gaseous pollutants enhance the inflammatory response to allergens, human data on whether combustion particulates enhance the inflammatory response to allergen are limited. Therefore, we conducted a human experimental study to investigate whether combustion particulates enhance the inflammatory response to aeroallergens. "Enhancement" refers to a greater-than-additive response when combustion particulates are delivered with allergen, compared with the responses when particulates and allergen are delivered alone. Eight subjects, five atopic and three nonatopic, participated in three randomized exposure-challenge sessions at least 2 weeks apart (i.e., clean air followed by allergen, particles followed by no allergen, or particles followed by allergen). Each session consisted of nasal exposure to combustion particles (target concentration of 1.0 mg/m3) or clean air for 1 hr, followed 3 hr later by challenge with whole pollen grains or placebo. Nasal lavage was performed immediately before particle or clean air exposure, immediately after exposure, and 4, 18 and 42 hr after pollen challenge. Cell counts, differentials, and measurement of cytokines were performed on each nasal lavage. In atopic but not in nonatopic subjects, when allergen was preceded by particulates, there was a significant enhancement immediately after pollen challenge in nasal lavage leukocytes and neutrophils (29.7 × 103 cells/mL and 25.4 × 103 cells/mL, respectively). This represents a 143% and 130% enhancement, respectively. The enhanced response for interleukin-4 was 3.23 pg/mL (p = 0.06), a 395% enhancement. In atopic subjects there was evidence of an enhanced response when particulates, as compared to clean air, preceded the allergen challenge.
Kammann, EE & Wand, M 2003, 'Geoadditive models', Journal of the Royal Statistical Society Series C: Applied Statistics, vol. 52, no. 1, pp. 1-18.
Kim, JY, Hauser, R, Wand, M, Herrick, RF, Amarasiriwardena, CJ & Christiani, DC 2003, 'The association of expired nitric oxide with occupational particulate metal exposure', Environmental Research, vol. 93, no. 2, pp. 158-166.View/Download from: Publisher's site
Kim, JY, Hauser, R, Wand, M, Herrick, RF, Houk, RS, Aeschliman, DB, Woodin, A & Christiani, DC 2003, 'Association of expired nitric oxide with urinary metal concentrations in boilermakers exposed to residual oil fly ash', American Journal of Industrial Medicine, vol. 44, no. 5, pp. 458-466.
Kim, JY, Wand, M, Hauser, R, Mukherjee, S, Herrick, RF & Christiani, DC 2003, 'Association of expired nitric oxide with occupational particulate exposure', Environmental Health Perspectives, vol. 111, no. 4, pp. 472-477.View/Download from: Publisher's site
Kim, JY, Wand, MP, Hauser, R, Mukherjee, S, Herrick, RF & Christiani, DC 2003, 'Association of expired nitric oxide with occupational particulate exposure', Environmental Health Perspectives, vol. 111, no. 5, pp. 676-680.View/Download from: Publisher's site
Particulate air pollution has been associated with adverse respiratory health effects. This study assessed the utility of expired nitric oxide to detect acute airway responses to metal-containing fine particulates. Using a repeated-measures study design, we investigated the association between the fractional concentration of expired nitric oxide (FENO) and exposure to particulate matter with an aerodynamic mass median diameter of ≤ 2.5 μm (PM2.5) in boilermakers exposed to residual oil fly ash and metal fumes. Subjects were monitored for 5 days during boiler repair overhauls in 1999 (n = 20 or 2000 (n = 14). The Wilcoxon median baseline FENO was 10.6 ppb [95% confidence interval (CI): 9.1, 12.7] in 1999 and 7.4 ppb (95% CI: 6.7, 8.0) in 2000. The Wilcoxon median PM2.5 8-hr time-weighted average was 0.56 mg/m3 (95% CI: 0.37, 0.93) in 1999 and 0.86 mg/m3 (95% CI: 0.65, 1.07) in 2000. FENO levels during the work week were significantly lower than baseline FENO in 1999 (p < 0.001). A significant inverse exposure-response relationship between log-transformed FENO and the previous workday's PM2.5 concentration was found in 1999, after adjusting for smoking status, age, and sampling year. With each 1 mg/m3 incremental increase in PM2.5 exposure, log FENO decreased by 0.24 (95% CI: -0.38, -0.10) in 1999. The lack of an exposure-response relationship between PM2.5 exposure and FENO in 2000 could be attributable to exposure misclassification resulting from the use of respirators. In conclusion, occupational exposure to metal-containing fine particulates was associated with significant decreases in FENO in a survey of workers with limited respirator usage.
Wand, M 2003, 'Smoothing and mixed models', Computational Statistics, vol. 18, pp. 223-249.
Aerts, M, Claeskens, G & Wand, M 2002, 'Some theory for penalized spline generalized additive models', Journal of Statistical Planning and Inference, vol. 103, no. 1-2, pp. 455-470.View/Download from: Publisher's site
Betensky, R, Lindsey, J, Wand, M & Ryan, LM 2002, 'A local likelihood proportional hazards model for interval censored data', Statistics in Medicine, vol. 21, pp. 263-275.View/Download from: Publisher's site
We discuss the use of local likelihood methods to fit proportional hazards regression models to right and interval censored data. The assumed model allows for an arbitrary, smoothed baseline hazard on which a vector of covariates operates in a proportional manner, and thus produces an interpretable baseline hazard function along with estimates of global covariate effects. For estimation, we extend the modified EM algorithm suggested by Betensky, Lindsey, Ryan and Wand. We illustrate the method with data on times to deterioration of breast cosmeses and HIV-1 infection rates among haemophiliacs
This article proposes a new method for estimation of the hazard function from a set of censored failure time data, with a view to extending the general approach to more complicated models. The approach is based on a mixed model representation of penalized spline hazard estimators. One payoff is the automation of the smoothing parameter choice through restricted maximum likelihood. Another is the option to use standard mixed model software for automatic hazard estimation. © 2002 American Statistical Association, Institute of Mathematical Statistics, and Interface Foundation of North America.
French, JL, Kammann, EE & Wand, MP 2002, 'Erratum: Semiparametric nonlinear mixed-effects models and their application (Journal of the American Statistical Association (2002) 96 (1285-1288))', Journal of the American Statistical Association, vol. 97, no. 458, p. 661.View/Download from: Publisher's site
French, JL, Kammann, EE & Wand, MP 2002, 'Semi-parametric nonlinear mixed-effects models and their application (vol 96, pg 1285)', JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION, vol. 97, no. 458, pp. 661-661.
Kim, EY, Zeng, Q, Rawn, J, Wand, M, Young, AJ, Milford, EL, Mentzer, SJ & Greenes, RA 2002, 'Using a neural network with flow cytometry histograms to recognize cell surface protein binding patterns', AMIA 2002 SYMPOSIUM, PROCEEDINGS, pp. 380-384.
Werneck, GL, Costa, CH, Walker, AM, David, JR, Wand, M & Maguire, JH 2002, 'The urban spread of visceral leishmaniasis: Clues from spatial analysis', Epidemiology, vol. 13, no. 3, pp. 364-367.View/Download from: Publisher's site
Background. The pattern of spread of visceral leishmaniasis in Brazilian cities is poorly understood. Methods. We used geographic information systems and spatial statistics to evaluate the distribution of 1061 cases of visceral leishmaniasis in Teresina, Brazil, in 1993 through 1996. Results. A locally weighted (LOESS) regression model, which was fit as a smoothed function of spatial coordinates, demonstrated large-scale variation, with high incidence rates in pe- ripheral neighborhoods that bordered forest land and pastures. Moran's I indicated small-scale variation and clustering up to 300 m, roughly the flight range of the sand fly vector. Conclusions. Spatial analytical techniques can identify high- risk areas for targeting control interventions
Zeng, Q, Wand, M, Young, AJ, Rawn, J, Milford, EL, Mentzer, SJ & Greenes, RA 2002, 'Matching of flow-cytometry histograms using information theory in feature space', AMIA 2002 SYMPOSIUM, PROCEEDINGS, pp. 929-933.
Coull, BA, Schwartz, J & Wand, M 2001, 'Respiratory health and air pollution: additive mixed model analyses', Biostatistics, vol. 2, no. 3, pp. 337-349.
French, JL, Kammann, EE & Wand, MP 2001, 'Semiparametric nonlinear mixed-effects models and their applications - Comment', JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION, vol. 96, no. 456, pp. 1285-1288.
Howard, RJ 2001, 'Perspective: media coverage of emerging and re-emerging diseases behind the headlines', Statistics in Medicine, vol. 20, no. 9-10, pp. 1357-1361.View/Download from: Publisher's site
Mammen, E, Marron, JS, Turlach, BA & Wand, M 2001, 'A general projection framework for constrained smoothing', Statistical Science, vol. 16, no. 3, pp. 232-248.
Parise, H, Wand, M, Ruppert, D & Ryan, LM 2001, 'Incorporation of historical controls using semiparametric mixed models', Journal of the Royal Statistical Society Series C: Applied Statistics, vol. 50, no. 1, pp. 31-42.
Zeng, Q, Young, AJ, Boxwala, AA, Rawn, J, Long, W, Wand, M, Salganik, M, Milford, EL, Mentzer, SJ & Greenes, RA 2001, 'Molecular identification using flow cytometry histograms and information theory', JOURNAL OF THE AMERICAN MEDICAL INFORMATICS ASSOCIATION, pp. 776-780.
Moore, PE, Laporte, JD, Abraham, JH, Schwartzman, IN, Yandava, CN, Silverman, ES, Drazen, JM, Wand, M, Panettieri, R & Shore, SA 2000, 'Polymorphism of the beta(2)-adrenergic receptor gene and desensitization in human airway smooth muscle', American Journal of Respiratory and Critical Care Medicine, vol. 162, pp. 2117-2124.View/Download from: Publisher's site
Wechsler, ME, Grasemann, H, Deykin, A, Silverman, EK, Yandava, CN, Isreal, E, Wand, M & Drazen, JM 2000, 'Exhaled nitric oxide in patients with asthma: Association with NOS1 genotype', American Journal of Respiratory and Critical Care Medicine, vol. 162, pp. 2043-2047.View/Download from: Publisher's site
Zanobetti, A, Wand, M, Schwartz, J & Ryan, LM 2000, 'Generalized additive distributed lag models: quantifying mortality displacement', Biostatistics, vol. 1, no. 3, pp. 279-292.View/Download from: Publisher's site
We propose a smooth hazard estimator for interval-censored survival data using the method of local likelihood. The model is fit using a local EM algorithm. The estimator is more descriptive than traditional empirical estimates in regions of concentrated information and takes on a parametric flavor in regions of sparse information. We derive two different standard error estimates for the smooth curve, one based on asymptotic theory and the other on the bootstrap. We illustrate the local EM method for times to breast cosmesis deterioration (Finkelstein, 1986, Biometrics 42, 845-854) and for times to HIV-1 infection for individuals with hemophilia (Kroner et al., 1994, Journal of AIDS 7, 279-286). Our hazard estimates for each of these data sets show interesting structures that would not be found using a standard parametric hazard model or empirical survivorship estimates.
Brumback, BA, Ruppert, D & Wand, MP 1999, 'Variable selection and function estimation in additive nonparametric regression using a data-based prior - Comment', JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION, vol. 94, no. 447, pp. 794-797.View/Download from: Publisher's site
Gijbels, I, Pope, A & Wand, M 1999, 'Understanding exponential smoothing via kernel regression', Journal of The Royal Statistical Society Series B-methodological, vol. 61, no. 1, pp. 39-50.
Augustyns, I & Wand, M 1998, 'Bandwidth selection for local polynomial smoothing of multinomial data', Computational Statistics, vol. 13, no. 4, pp. 447-461.
A semiparametric version of the generalized linear model for regression response was developed by replacing the linear combination with nonparametric components. The generalized partially linear single-index models were formed by combining simpler, conventional models such as single-index and partially linear models. Furthermore, the asymptotic distributions of the linear combination involving unknown parameters and unknown function was obtained by using local linear methods.
Wand, M & Gutierrez, RG 1997, 'Exact risk approaches to smoothing parameter selection', Journal of Nonparametric Statistics, vol. 8, no. 4, pp. 337-354.
Gonzalez-Manteiga, W, Sanchez-Sellero, C & Wand, M 1996, 'Accuracy of binned kernel functional approximations', Computational Statistics and Data Analysis, vol. 22, no. 1, pp. 1-16.View/Download from: Publisher's site
Virtually all common bandwidth selection algorithms are based on a certain type of kernel functional estimator. Such estimators can be computationally very expensive, so in practice they are often replaced by fast binned approximations. This is especially worthwhile when the bandwidth selection method involves iteration. Results for the accuracy of these approximations are derived and then used to provide an understanding of the number of binning grid points required to achieve a given level of accuracy. Our results apply to both univariate and multivariate settings. Multivariate contexts are of particular interest since the cost due to having a higher number of grid points can be quite significant.
Ruppert, D, Sheather, SJ & Wand, MP 1996, 'An effective bandwidth selector for local least squares regression (vol 90 pg 1257, 1995)', JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION, vol. 91, no. 435, pp. 1380-1380.
Turlach, BA & Wand, M 1996, 'Fast computation of auxiliary quantities in local polynomial regression', Journal of Computational and Graphical Statistics, vol. 5, no. 4, pp. 337-350.
We investigate the extension of binning methodology to fast computation of several auxiliary quantities that arise in local polynomial smoothing. Examples include degrees of freedom measures, cross-validation functions, variance estimates, and exact measures of error. It is shown that the computational effort required for such approximations is of the same order of magnitude as that required for a binned local polynomial smooth.
Aldershof, B, Marron, JS, Park, BU & Wand, M 1995, 'Facts about the gaussian probability density function', Applicable Analysis:, vol. 59, no. 1, pp. 289-306.
Fan, J, Heckman, NE & Wand, M 1995, 'Local polynomial kernel regression for generalized linear models and quasi-likelihood functions', Journal of the American Statistical Association, vol. 90, no. 429, pp. 141-150.View/Download from: Publisher's site
Fan, J, Heckman, NE & Wand, MP 1995, 'Local polynomial kernel regression for generalized linear models and quasi-likelihood functions', Journal of the American Statistical Association, vol. 90, no. 429, pp. 141-150.View/Download from: Publisher's site
We investigate the extension of the nonparametric regression technique of local polynomial fitting with a kernel weight to generalized linear models and quasi-likelihood contexts. In the ordinary regression case, local polynomial fitting has been seen to have several appealing features in terms of intuitive and mathematical simplicity. One noteworthy feature is the better performance near the boundaries compared to the traditional kernel regression estimators. These properties are shown to carry over to generalized linear model and quasi-likelihood settings. We also derive the asymptotic distributions of the proposed class of estimators that allow for straightforward interpretation and extensions of state-of-the-art bandwidth selection methods. © 1995 Taylor & Francis Group, LLC.
Herrmann, E, Wand, M, Engel, J & Gasser, T 1995, 'A bandwidth selector for bivariate kernel regression', Journal of The Royal Statistical Society Series B-methodological, vol. 57, no. 1, pp. 171-180.
Ruppert, D, Sheather, SJ & Wand, M 1995, 'An effective bandwidth selector for local least squares regression', Journal of the American Statistical Association, vol. 90, no. 432, pp. 1257-1270.
Local least squares kernel regression provides an appealing solution to the nonparametric regression, or "scatterplot smoothing," problem, as demonstrated by Fan, for example. The practical implementation of any scatterplot smoother is greatly enhanced by the availability of a reliable rule for automatic selection of the smoothing parameter. In this article we apply the ideas of plug-in bandwidth selection to develop strategies for choosing the smoothing parameter of local linear squares kernel estimators. Our results are applicable to odd-degree local polynomial fits and can be extended to other settings, such as derivative estimation and multiple nonparametric regression. An implementation in the important case of local linear fits with univariate predictors is shown to perform well in practice. A by-product of our work is the development of a class of nonparametric variance estimators, based on local least squares ideas, and plug-in rules for their implementation.
Ruppert, D, Sheather, SJ & Wand, MP 1995, 'An effective bandwidth selector for local least squares regression', JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION, vol. 90, no. 432, pp. 1257-1270.View/Download from: Publisher's site
Wand, M 1994, 'Fast computation of multivariate kernel estimators', Journal of Computational and Graphical Statistics, vol. 3, no. 4, pp. 433-445.
Wand, M & Jones, MC 1994, 'Multivariate plug-in bandwidth selection', Computational Statistics, vol. 9, pp. 97-116.
Multivariate extensions of binning techniques for fast computation of kernel estimators are described and examined. Several questions arising from this multivariate extension are addressed. The choice of binning rule is discussed, and it is demonstrated that linear binning leads to substantial accuracy improvements over simple binning. An investigation into the most appropriate means of computing the multivariate discrete convolutions required for binned kernel estimators is also given. The results of an empirical study indicate that, in multivariate settings, the fast Fourier transform offers considerable time savings compared to direct calculation of convolutions. © 1994, American Statistical Association, Institute of Mathematical Statistics, and Interface Foundation of North America.
Devroye, L & Wand, M 1993, 'On the influence of the density on the kernel estimate', Statistics, vol. 24, pp. 215-233.
We introduce a universal functional Q(f) which measures the difficulty a given density f poses to the standard kernel density estimate if one uses the optimal smoothing factor, The functional is well-defined (but possibly infinite) for all densities, regardless of their smoothness or tail properties. It is proportional to the limit of n2/5E ∫ |fn — f| where fnis the optimal kernel estimate. This paper settles some questions left unanswered in Devroye and Györfi (1985) and Hall and Wand (1988). © 1993, Taylor & Francis Group, LLC. All rights reserved.
Wand, M & Jones, MC 1993, 'Comparison of smoothing parameterizations in bivariate kernel density estimation', Journal of the American Statistical Association, vol. 88, no. 422, pp. 520-528.View/Download from: Publisher's site
Wand, MP & Jones, MC 1993, 'Comparison of smoothing parameterizations in bivariate kernel density estimation', Journal of the American Statistical Association, vol. 88, no. 422, pp. 520-528.View/Download from: Publisher's site
The basic kernel density estimator in one dimension has a single smoothing parameter, usually referred to as the bandwidth. For higher dimensions, however, there are several options for smoothing parameterization of the kernel estimator. For the bivariate case, there can be between one and three independent smoothing parameters in the estimator, which leads to a flexibility versus complexity trade-off when using this estimator in practice. In this article the performances of the different possible smoothing parameterizations are compared, using both the asymptotic and exact mean integrated squared error. Our results show that it is important to have independent smoothing parameters for each of the coordinate directions. Although this is enough for many situations, for densities with high amounts of curvature in directions different to those of the coordinate axes, substantial gains can be made by allowing the kernel mass to have arbitrary orientations. The "sphering" approaches to choosing this orientation are shown to be detrimental in general, however. © 1993 Taylor & Francis Group, LLC.
Ruppert, D & Wand, M 1992, 'Correcting for kurtosis in density estimation', Australian Journal of Statistics, vol. 34`, pp. 19-29.
Wand, M 1992, 'Error analyses for general multivariate kernel estimators', Journal of Nonparametric Statistics, vol. 2, pp. 1-15.
Carroll, RJ & Wand, M 1991, 'Semiparametric estimation in logistic measurement error models', Journal of The Royal Statistical Society Series B-methodological, vol. 53, no. 3, pp. 573-585.
Scott, DW & Wand, M 1991, 'Feasibility of multivariate density estimates', Biometrika, vol. 78, no. 1, pp. 197-205.
For the density estimation problem the global window width kernel density estimator does not perform well when the underlying density has features that require different amounts of smoothing at different locations. In this article we propose to transform the data with the intention that a global window width is more appropriate for the density of the transformed data. The density estimate of the original data is the "back-transform" by change of variables of the global window width estimate of the transformed data's density. We explore choosing the transformation from suitable parametric families. Data-based selection rules for the choice of transformations and the window width are discussed. Application to real and simulated data demonstrates the usefulness of our proposals. © 1991 Taylor & Francis Group, LLC.
WAND, MP, MARRON, JS & RUPPERT, D 1991, 'TRANSFORMATIONS IN DENSITY-ESTIMATION - REJOINDER', JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION, vol. 86, no. 414, pp. 360-361.View/Download from: Publisher's site
Hardle, W, Marron, JS & Wand, M 1990, 'Bandwidth choice for density derivatives', Journal of The Royal Statistical Society Series B-methodological, vol. 52, no. 1, pp. 223-232.
Wand, M 1990, 'On exact L1 rates of convergence in non-parametric kernel regression', Scandinavian Journal of Statistics, vol. 17, no. 3, pp. 251-256.
Maestrini, L & Wand, MP, 'The Inverse G-Wishart Distribution and Variational Message Passing'.
Message passing on a factor graph is a powerful paradigm for the coding of
approximate inference algorithms for arbitrarily graphical large models. The
notion of a factor graph fragment allows for compartmentalization of algebra
and computer code. We show that the Inverse G-Wishart family of distributions
enables fundamental variational message passing factor graph fragments to be
expressed elegantly and succinctly. Such fragments arise in models for which
approximate inference concerning covariance matrix or variance parameters is
made, and are ubiquitous in contemporary statistics and machine learning.
Menictas, M, Credico, GD & Wand, MP, 'Streamlined Variational Inference for Linear Mixed Models with Crossed Random Effects'.
We derive streamlined mean field variational Bayes algorithms for fitting
linear mixed models with crossed random effects. In the most general situation,
where the dimensions of the crossed groups are arbitrarily large, streamlining
is hindered by lack of sparseness in the underlying least squares system.
Because of this fact we also consider a hierarchy of relaxations of the mean
field product restriction. The least stringent product restriction delivers a
high degree of inferential accuracy. However, this accuracy must be mitigated
against its higher storage and computing demands. Faster sparse storage and
computing alternatives are also provided, but come with the price of diminished
inferential accuracy. This article provides full algorithmic details of three
variational inference strategies, presents detailed empirical results on their
pros and cons and, thus, guides the users on their choice of variational
inference approach depending on the problem size and computing resources.
Nolan, TH & Wand, MP, 'Solutions to Multilevel Sparse Matrix Problems'.
We define and solve classes of sparse matrix problems that arise in
multilevel modeling and data analysis. The classes are indexed by the number of
nested units, with two-level problems corresponding to the common situation in
which data on level 1 units are grouped within a two-level structure. We
provide full solutions for two-level and three-level problems and their
derivations provide blueprints for the challenging, albeit rarer in
applications, higher level versions of the problem. Whilst our linear system
solutions are a concise recasting of existing results, our matrix inverse
sub-block results are novel and facilitate streamlined computation of standard
errors in frequentist inference as well as allowing streamlined mean field
variational Bayesian inference for models containing higher level random
We define and solve classes of sparse matrix problems that arise in multilevel modelling and data analysis. The classes are indexed by the number of nested units, with two-level problems corresponding to the common situation, in which data on level-1 units are grouped within a two-level structure. We provide full solutions for two-level and three-level problems, and their derivations provide blueprints for the challenging, albeit rarer in applications, higher-level versions of the problem. While our linear system solutions are a concise recasting of existing results, our matrix inverse sub-block results are novel and facilitate streamlined computation of standard errors in frequentist inference as well as allowing streamlined mean field variational Bayesian inference for models containing higher-level random effects.
Nolan, TH, Menictas, M & Wand, MP, 'Streamlined Computing for Variational Inference with Higher Level Random Effects'.
We derive and present explicit algorithms to facilitate streamlined computing
for variational inference for models containing higher level random effects.
Existing literature, such as Lee and Wand (2016), is such that streamlined
variational inference is restricted to mean field variational Bayes algorithms
for two-level random effects models. Here we provide the following extensions:
(1) explicit Gaussian response mean field variational Bayes algorithms for
three-level models, (2) explicit algorithms for the alternative variational
message passing approach in the case of two-level and three-level models, and
(3) an explanation of how arbitrarily high levels of nesting can be handled
based on the recently published matrix algebraic results of the authors. A
pay-off from (2) is simple extension to non-Gaussian response models. In
summary, we remove barriers for streamlining variational inference algorithms
based on either the mean field variational Bayes approach or the variational
message passing approach when higher level random effects are present.
Wand, MP & Yu, JCF, 'Density Estimation via Bayesian Inference Engines'.
We explain how effective automatic probability density function estimates can
be constructed using contemporary Bayesian inference engines such as those
based on no-U-turn sampling and expectation propagation. Extensive simulation
studies demonstrate that the proposed density estimates have excellent
comparative performance and scale well to very large sample sizes due a binning
strategy. Moreover, the approach is fully Bayesian and all estimates are
accompanied by pointwise credible intervals. An accompanying package in the R
language facilitates easy use of the new density estimates.
Michaelson, G, Roughan, M, Tuke, J, Wand, MP & Bush, R 2018, 'Rasch analysis of HTTPS reachability', 2018 IFIP NETWORKING CONFERENCE (IFIP NETWORKING) AND WORKSHOPS, 17th IFIP Networking Conference (IFIP Networking), IEEE, Univ Zurich, Zurich, SWITZERLAND, pp. 1-9.
Naumann, U, Wand, MP & George, L 2011, 'AUTOMATION IN HIGH-CONTENT FLOW CYTOMETRY SCREENING', CYTOMETRY PART B-CLINICAL CYTOMETRY, 26th Annual Meeting of the International-Clinical-Cytometry-Society, WILEY-BLACKWELL, Portland, OR, pp. 391-391.
Nevillea, SE & Wand, M 2011, 'Generalised Extreme Value geoadditive model analysis via variational Bayes', Procedia Environmental Sciences, International Conference on Spatial Statistics - Mapping Global Change, Elsevier, Enschede, pp. 8-13.View/Download from: Publisher's site
We devise a variationalBayes algorithm for fast approximate inference in Bayesian GeneralizedExtremeValue additive modelanalysis. Such models are useful for flexibly assessing the impact of continuous predictor variables on sample extremes. The new methodology allows large Bayesian models to be fitted and assessed without the significant computing costs of Monte Carlo methods
Massaro, AF, Coulston, EL, Foster, J, Devkin, A, Yim, E, Mazzella, L, Wand, MP, Salganik, MP, Israel, E & Drazen, JM 1999, 'Diurnal variation in expired nitric oxide (NO) in asthmatics.', AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE, AMER THORACIC SOC, pp. A860-A860.
Opsomer, JD, Ruppert, D, Wand, MP, Holst, U & Hoessjer, O 1998, 'Kriging with nonparametric variance function estimation', DIMENSION REDUCTION, COMPUTATIONAL COMPLEXITY AND INFORMATION, 30th Symposium on Interface Between Computing Science and Statistics, INTERFACE FOUNDATION NORTH AMERICA, MINNEAPOLIS, MN, pp. 111-111.
Hall, P, Johnstone, IM, Ormerod, JT, Wand, MP & Yu, JCF 2020, 'Fast and Accurate Binary Response Mixed Model Analysis via Expectation Propagation'.
Expectation propagation is a general prescription for approximation of
integrals in statistical inference problems. Its literature is mainly concerned
with Bayesian inference scenarios. However, expectation propagation can also be
used to approximate integrals arising in frequentist statistical inference. We
focus on likelihood-based inference for binary response mixed models and show
that fast and accurate quadrature-free inference can be realized for the probit
link case with multivariate random effects and higher levels of nesting. The
approach is supported by asymptotic theory in which expectation propagation is
seen to provide consistent estimation of the exact likelihood surface.
Numerical studies reveal the availability of fast, highly accurate and scalable
methodology for binary mixed model analysis.
Chen, WY & Wand, MP 2018, 'Factor graph fragmentization of expectation propagation'.
Expectation propagation is a general approach to fast approximate inference
for graphical models. The existing literature treats models separately when it
comes to deriving and coding expectation propagation inference algorithms. This
comes at the cost of similar, long-winded algebraic steps being repeated and
slowing down algorithmic development. We demonstrate how factor graph
fragmentization can overcome this impediment. This involves adoption of the
message passing on a factor graph approach to expectation propagation and
identification of factor graph sub-graphs, which we call fragments, that are
common to wide classes of models. Key fragments and their corresponding
messages are catalogued which means that their algebra does not need to be
repeated. This allows compartmentalization of coding and efficient software