arXiv subjects

Mike West

Publications and source records attributed to Mike West.

At least 37 records · Page 2Linked to original sources

arXiv2020

Bayesian Computation in Dynamic Latent Factor Models

Bayesian computation for filtering and forecasting analysis is developed for a broad class of dynamic models. The ability to scale-up such analyses in non-Gaussian, nonlinear multivariate time series models is advanced through the introduction of a novel copula construction in sequential filtering of coupled sets of dynamic generalized linear models. The new copula approach is integrated into recently introduced multiscale models in which univariate time series are coupled via nonlinear forms involving dynamic latent factors representing cross-series relationships. The resulting methodology offers dramatic speed-up in online Bayesian computations for sequential filtering and forecasting in this broad, flexible class of multivariate models. Two examples in nonlinear models for very heterogeneous time series of non-negative counts demonstrate massive computational efficiencies relative to existing simulation-based methods, while defining similar filtering and forecasting outcomes.

stat.ME↗

arXiv2019

Bayesian forecasting of multivariate time series: Scalability, structure uncertainty and decisions

I overview recent research advances in Bayesian state-space modeling of multivariate time series. A main focus is on the decouple/recouple concept that enables application of state-space models to increasingly large-scale data, applying to continuous or discrete time series outcomes. The scope includes large-scale dynamic graphical models for forecasting and multivariate volatility analysis in areas such as economics and finance, multi-scale approaches for forecasting discrete/count time series in areas such as commercial sales and demand forecasting, and dynamic network flow models for areas including internet traffic monitoring. In applications, explicit forecasting, monitoring and decision goals are paramount and should factor into model assessment and comparison, a perspective that is highlighted.

stat.ME↗

arXiv2019

Bayesian Dynamic Modeling and Monitoring of Network Flows

In the context of a motivating study of dynamic network flow data on a large-scale e-commerce web site, we develop Bayesian models for on-line/sequential analysis for monitoring and adapting to changes reflected in node-node traffic. For large-scale networks, we customize core Bayesian time series analysis methods using dynamic generalized linear models (DGLMs). These are integrated into the context of multivariate networks using the concept of decouple/recouple that was recently introduced in multivariate time series. This method enables flexible dynamic modeling of flows on large-scale networks and exploitation of partial parallelization of analysis while maintaining coherence with an over-arching multivariate dynamic flow model. This approach is anchored in a case-study on internet data, with flows of visitors to a commercial news web site defining a long time series of node-node counts on over 56,000 node pairs. Central questions include characterizing inherent stochasticity in traffic patterns, understanding node-node interactions, adapting to dynamic changes in flows and allowing for sensitive monitoring to flag anomalies. The methodology of dynamic network DGLMs applies to many dynamic network flow studies.

stat.ME↗

arXiv2018

Variable Prioritization in Nonlinear Black Box Methods: A Genetic Association Case Study

The central aim in this paper is to address variable selection questions in nonlinear and nonparametric regression. Motivated by statistical genetics, where nonlinear interactions are of particular interest, we introduce a novel and interpretable way to summarize the relative importance of predictor variables. Methodologically, we develop the "RelATive cEntrality" (RATE) measure to prioritize candidate genetic variants that are not just marginally important, but whose associations also stem from significant covarying relationships with other variants in the data. We illustrate RATE through Bayesian Gaussian process regression, but the methodological innovations apply to other "black box" methods. It is known that nonlinear models often exhibit greater predictive accuracy than linear models, particularly for phenotypes generated by complex genetic architectures. With detailed simulations and two real data association mapping studies, we show that applying RATE enables an explanation for this improved performance.

stat.ME↗

arXiv2018

Probabilistic forecasting of heterogeneous consumer transaction-sales time series

We present new Bayesian methodology for consumer sales forecasting. With a focus on multi-step ahead forecasting of daily sales of many supermarket items, we adapt dynamic count mixture models to forecast individual customer transactions, and introduce novel dynamic binary cascade models for predicting counts of items per transaction. These transactions-sales models can incorporate time-varying trend, seasonal, price, promotion, random effects and other outlet-specific predictors for individual items. Sequential Bayesian analysis involves fast, parallel filtering on sets of decoupled items and is adaptable across items that may exhibit widely varying characteristics. A multi-scale approach enables information sharing across items with related patterns over time to improve prediction while maintaining scalability to many items. A motivating case study in many-item, multi-period, multi-step ahead supermarket sales forecasting provides examples that demonstrate improved forecast accuracy in multiple metrics, and illustrates the benefits of full probabilistic models for forecast accuracy evaluation and comparison. Keywords: Bayesian forecasting; decouple/recouple; dynamic binary cascade; forecast calibration; intermittent demand; multi-scale forecasting; predicting rare events; sales per transaction; supermarket sales forecasting

stat.ME↗

arXiv2018

Multivariate Bayesian Predictive Synthesis in Macroeconomic Forecasting

We develop the methodology and a detailed case study in use of a class of Bayesian predictive synthesis (BPS) models for multivariate time series forecasting. This extends the recently introduced foundational framework of BPS to the multivariate setting, with detailed application in the topical and challenging context of multi-step macroeconomic forecasting in a monetary policy setting. BPS evaluates-- sequentially and adaptively over time-- varying forecast biases and facets of miscalibration of individual forecast densities, and-- critically-- of time-varying inter-dependencies among them over multiple series. We develop new BPS methodology for a specific subclass of the dynamic multivariate latent factor models implied by BPS theory. Structured dynamic latent factor BPS is here motivated by the application context-- sequential forecasting of multiple US macroeconomic time series with forecasts generated from several traditional econometric time series models. The case study highlights the potential of BPS to improve of forecasts of multiple series at multiple forecast horizons, and its use in learning dynamic relationships among forecasting models or agents.

stat.ME↗

arXiv2018

Bayesian forecasting of many count-valued time series

This paper develops forecasting methodology and application of new classes of dynamic models for time series of non-negative counts. Novel univariate models synthesise dynamic generalized linear models for binary and conditionally Poisson time series, with dynamic random effects for over-dispersion. These models allow use of dynamic covariates in both binary and non-zero count components. Sequential Bayesian analysis allows fast, parallel analysis of sets of decoupled time series. New multivariate models then enable information sharing in contexts when data at a more highly aggregated level provide more incisive inferences on shared patterns such as trends and seasonality. A novel multi-scale approach-- one new example of the concept of decouple/recouple in time series-- enables information sharing across series. This incorporates cross-series linkages while insulating parallel estimation of univariate models, hence enables scalability in the number of series. The major motivating context is supermarket sales forecasting. Detailed examples drawn from a case study in multi-step forecasting of sales of a number of related items showcase forecasting of multiple series, with discussion of forecast accuracy metrics and broader questions of probabilistic forecast accuracy assessment.

stat.ME↗

arXiv2016

Scalable Bayesian modeling, monitoring and analysis of dynamic network flow data

Traffic flow count data in networks arise in many applications, such as automobile or aviation transportation, certain directed social network contexts, and Internet studies. Using an example of Internet browser traffic flow through site-segments of an international news website, we present Bayesian analyses of two linked classes of models which, in tandem, allow fast, scalable and interpretable Bayesian inference. We first develop flexible state-space models for streaming count data, able to adaptively characterize and quantify network dynamics efficiently in real-time. We then use these models as emulators of more structured, time-varying gravity models that allow formal dissection of network dynamics. This yields interpretable inferences on traffic flow characteristics, and on dynamics in interactions among network nodes. Bayesian monitoring theory defines a strategy for sequential model assessment and adaptation in cases when network flow data deviates from model-based predictions. Exploratory and sequential monitoring analyses of evolving traffic on a network of web site-segments in e-commerce demonstrate the utility of this coupled Bayesian emulation approach to analysis of streaming network count data.

stat.ME↗

arXiv2016

Bayesian emulation for optimization in multi-step portfolio decisions

We discuss the Bayesian emulation approach to computational solution of multi-step portfolio studies in financial time series. "Bayesian emulation for decisions" involves mapping the technical structure of a decision analysis problem to that of Bayesian inference in a purely synthetic "emulating" statistical model. This provides access to standard posterior analytic, simulation and optimization methods that yield indirect solutions of the decision problem. We develop this in time series portfolio analysis using classes of economically and psychologically relevant multi-step ahead portfolio utility functions. Studies with multivariate currency, commodity and stock index time series illustrate the approach and show some of the practical utility and benefits of the Bayesian emulation methodology.

stat.ME↗

arXiv2016

Bayesian analysis of immune response dynamics with sparse time series data

In vaccine development, the temporal profiles of relative abundance of subtypes of immune cells (T-cells) is key to understanding vaccine efficacy. Complex and expensive experimental studies generate very sparse time series data on this immune response. Fitting multi-parameter dynamic models of the immune response dynamics-- central to evaluating mechanisms underlying vaccine efficacy-- is challenged by data sparsity. The research reported here addresses this challenge. For HIV/SIV vaccine studies in macaques, we: (a) introduce novel dynamic models of progression of cellular populations over time with relevant, time-delayed components reflecting the vaccine response; (b) define an effective Bayesian model fitting strategy that couples Markov chain Monte Carlo (MCMC) with Approximate Bayesian Computation (ABC)-- building on the complementary strengths of the two approaches, neither of which is effective alone; (c) explore questions of information content in the sparse time series for each of the model parameters, linking into experimental design and model simplification for future experiments; and (d) develop, apply and compare the analysis with samples from a recent HIV/SIV experiment, with novel insights and conclusions about the progressive response to the vaccine, and how this varies across subjects.

stat.ME↗

arXiv2016

Dynamic dependence networks: Financial time series forecasting and portfolio decisions (with discussion)

We discuss Bayesian forecasting of increasingly high-dimensional time series, a key area of application of stochastic dynamic models in the financial industry and allied areas of business. Novel state-space models characterizing sparse patterns of dependence among multiple time series extend existing multivariate volatility models to enable scaling to higher numbers of individual time series. The theory of these "dynamic dependence network" models shows how the individual series can be "decoupled" for sequential analysis, and then "recoupled" for applied forecasting and decision analysis. Decoupling allows fast, efficient analysis of each of the series in individual univariate models that are linked-- for later recoupling-- through a theoretical multivariate volatility structure defined by a sparse underlying graphical model. Computational advances are especially significant in connection with model uncertainty about the sparsity patterns among series that define this graphical model; Bayesian model averaging using discounting of historical information builds substantially on this computational advance. An extensive, detailed case study showcases the use of these models, and the improvements in forecasting and financial portfolio investment decisions that are achievable. Using a long series of daily international currency, stock indices and commodity prices, the case study includes evaluations of multi-day forecasts and Bayesian portfolio analysis with a variety of practical utility functions, as well as comparisons against commodity trading advisor benchmarks.

stat.ME↗

arXiv2016

Models of random sparse eigenmatrices matrices and Bayesian analysis of multivariate structure

We discuss probabilistic models of random covariance structures defined by distributions over sparse eigenmatrices. The decomposition of orthogonal matrices in terms of Givens rotations defines a natural, interpretable framework for defining distributions on sparsity structure of random eigenmatrices. We explore theoretical aspects and implications for conditional independence structures arising in multivariate Gaussian models, and discuss connections with sparse PCA, factor analysis and Gaussian graphical models. Methodology includes model-based exploratory data analysis and Bayesian analysis via reversible jump Markov chain Monte Carlo. A simulation study examines the ability to identify sparse multivariate structures compared to the benchmark graphical modelling approach. Extensions to multivariate normal mixture models with additional measurement errors move into the framework of latent structure analysis of broad practical interest. We explore the implications and utility of the new models with summaries of a detailed applied study of a 20-dimensional breast cancer genomics data set.

stat.ME↗

arXiv2016

Bayesian forecasting and scalable multivariate volatility analysis using simultaneous graphical dynamic models

The recently introduced class of simultaneous graphical dynamic linear models (SGDLMs) defines an ability to scale on-line Bayesian analysis and forecasting to higher-dimensional time series. This paper advances the methodology of SGDLMs, developing and embedding a novel, adaptive method of simultaneous predictor selection in forward filtering for on-line learning and forecasting. The advances include developments in Bayesian computation for scalability, and a case study in exploring the resulting potential for improved short-term forecasting of large-scale volatility matrices. A case study concerns financial forecasting and portfolio optimization with a 400-dimensional series of daily stock prices. Analysis shows that the SGDLM forecasts volatilities and co-volatilities well, making it ideally suited to contributing to quantitative investment strategies to improve portfolio returns. We also identify performance metrics linked to the sequential Bayesian filtering analysis that turn out to define a leading indicator of increased financial market stresses, comparable to but leading the standard St. Louis Fed Financial Stress Index (STLFSI) measure. Parallel computation using GPU implementations substantially advance the ability to fit and use these models.

stat.AP↗

arXiv2016

Dynamics and sparsity in latent threshold factor models: A study in multivariate EEG signal processing

We discuss Bayesian analysis of multivariate time series with dynamic factor models that exploit time-adaptive sparsity in model parametrizations via the latent threshold approach. One central focus is on the transfer responses of multiple interrelated series to underlying, dynamic latent factor processes. Structured priors on model hyper-parameters are key to the efficacy of dynamic latent thresholding, and MCMC-based computation enables model fitting and analysis. A detailed case study of electroencephalographic (EEG) data from experimental psychiatry highlights the use of latent threshold extensions of time-varying vector autoregressive and factor models. This study explores a class of dynamic transfer response factor models, extending prior Bayesian modeling of multiple EEG series and highlighting the practical utility of the latent thresholding concept in multivariate, non-stationary time series analysis.

stat.AP↗

arXiv2015

Sequential Monte Carlo with Adaptive Weights for Approximate Bayesian Computation

Methods of approximate Bayesian computation (ABC) are increasingly used for analysis of complex models. A major challenge for ABC is over-coming the often inherent problem of high rejection rates in the accept/reject methods based on prior:predictive sampling. A number of recent developments aim to address this with extensions based on sequential Monte Carlo (SMC) strategies. We build on this here, introducing an ABC SMC method that uses data-based adaptive weights. This easily implemented and computationally trivial extension of ABC SMC can very substantially improve acceptance rates, as is demonstrated in a series of examples with simulated and real data sets, including a currently topical example from dynamic modelling in systems biology applications.

stat.CO↗

arXiv2011

Autoregressive Models for Variance Matrices: Stationary Inverse Wishart Processes

We introduce and explore a new class of stationary time series models for variance matrices based on a constructive definition exploiting inverse Wishart distribution theory. The main class of models explored is a novel class of stationary, first-order autoregressive (AR) processes on the cone of positive semi-definite matrices. Aspects of the theory and structure of these new models for multivariate "volatility" processes are described in detail and exemplified. We then develop approaches to model fitting via Bayesian simulation-based computations, creating a custom filtering method that relies on an efficient innovations sampler. An example is then provided in analysis of a multivariate electroencephalogram (EEG) time series in neurological studies. We conclude by discussing potential further developments of higher-order AR models and a number of connections with prior approaches.

stat.ME↗

arXiv2010

An integrative analysis of cancer gene expression studies using Bayesian latent factor modeling

We present an applied study in cancer genomics for integrating data and inferences from laboratory experiments on cancer cell lines with observational data obtained from human breast cancer studies. The biological focus is on improving understanding of transcriptional responses of tumors to changes in the pH level of the cellular microenvironment. The statistical focus is on connecting experimentally defined biomarkers of such responses to clinical outcome in observational studies of breast cancer patients. Our analysis exemplifies a general strategy for accomplishing this kind of integration across contexts. The statistical methodologies employed here draw heavily on Bayesian sparse factor models for identifying, modularizing and correlating with clinical outcome these signatures of aggregate changes in gene expression. By projecting patterns of biological response linked to specific experimental interventions into observational studies where such responses may be evidenced via variation in gene expression across samples, we are able to define biomarkers of clinically relevant physiological states and outcomes that are rooted in the biology of the original experiment. Through this approach we identify microenvironment-related prognostic factors capable of predicting long term survival in two independent breast cancer datasets. These results suggest possible directions for future laboratory studies, as well as indicate the potential for therapeutic advances though targeted disruption of specific pathway components.

stat.AP↗

arXiv2007

The Use of Unlabeled Data in Predictive Modeling

The incorporation of unlabeled data in regression and classification analysis is an increasing focus of the applied statistics and machine learning literatures, with a number of recent examples demonstrating the potential for unlabeled data to contribute to improved predictive accuracy. The statistical basis for this semisupervised analysis does not appear to have been well delineated; as a result, the underlying theory and rationale may be underappreciated, especially by nonstatisticians. There is also room for statisticians to become more fully engaged in the vigorous research in this important area of intersection of the statistical and computer sciences. Much of the theoretical work in the literature has focused, for example, on geometric and structural properties of the unlabeled data in the context of particular algorithms, rather than probabilistic and statistical questions. This paper overviews the fundamental statistical foundations for predictive modeling and the general questions associated with unlabeled data, highlighting the relevance of venerable concepts of sampling design and prior specification. This theory, illustrated with a series of central illustrative examples and two substantial real data analyses, shows precisely when, why and how unlabeled data matter.

stat.ME↗