Nonparametric Bayesian Modeling for Stochastic Order

In comparing two populations, sometimes a model incorporating stochastic order is desired. Customarily, such modeling is done parametrically. The objective of this paper is to formulate nonparametric (possibly semiparametric) stochastic order specifications providing richer, more flexible modeling. We adopt a fully Bayesian approach using Dirichlet process mixing. An attractive feature of the Bayesian approach is that full inference is available regarding the population distributions. Prior information can conveniently be incorporated. Also, prior stochastic order is preserved to the posterior analysis. Apart from the two sample setting, the approach handles the matched pairs problem, the k-sample slippage problem, ordered ANOVA and ordered regression models. We illustrate by comparing two rather small samples, one of diabetic men, the other of diabetic women. Measurements are of androstenedione levels. Males are anticipated to produce levels which will tend to be higher than those of females.     

Efficient Data Augmentation for Fitting Stochastic Epidemic Models to Prevalence Data

Stochastic epidemic models describe the dynamics of an epidemic as a disease spreads through a population. Typically, only a fraction of cases are observed at a set of discrete times. The absence of complete information about the time evolution of an epidemic gives rise to a complicated latent variable problem in which the state space size of the epidemic grows large as the population size increases. This makes analytically integrating over the missing data infeasible for populations of even moderate size. We present a data augmentation Markov chain Monte Carlo (MCMC) framework for Bayesian estimation of stochastic epidemic model parameters, in which measurements are augmented with subject-level disease histories. In our MCMC algorithm, we propose each new subject-level path, conditional on the data, using a time-inhomogenous continuous-time Markov process with rates determined by the infection histories of other individuals. The method is general, and may be applied to a broad class of epidemic models with only minimal modifications to the model dynamics and/or emission distribution. We present our algorithm in the context of multiple stochastic epidemic models in which the data are binomially sampled prevalence counts, and apply our method to data from an outbreak of influenza in a British boarding school. Supplementary material for this article is available online.     

EFFECTS OF INDIVIDUAL HETEROGENEITY IN ESTIMATING THE PERSISTENCE OF SMALL POPULATIONS

ABSTRACT. Population viability models are commonly used to estimate the probability of persistence of small, threatened, or endangered populations. Demographic, temporal, spatial, and individual heterogeneity are important factors affecting the probability of persistence of small populations. Because stochastic process are intractable analytically (Lud-wig [1996]), computer simulation models are often used for estimating population viability via numerical techniques. Although demographic, spatial, and temporal stochasticity have been incorporated into some population viability models, individual heterogeneity has not been included. In this paper we include individual heterogeneity in a simulation model and examine probabilities of population persistence at different levels of heterogeneity and population size. Individual heterogeneity may increase the probability of persistence of small populations. The mechanism for the extension in persistence may be explained by natural selection. Genotypes persisting through a decline may be those that survive better under the conditions causing the decline. These individuals that survive and reproduce in the face of adverse conditions may extend the probability that a small population persists.     

Modeling Variability Order: A Semiparametric Bayesian Approach

In comparing two populations, sometimes a model incorporating a certain probability order is desired. In this setting, Bayesian modeling is attractive since a probability order restriction imposed a priori on the population distributions is retained a posteriori. Extending the work in Gelfand and Kottas (2001) for stochastic order specifications, we formulate modeling for distributions ordered in variability. We work with Dirichlet process mixtures resulting in a fully Bayesian semiparametric approach. The details for simulation-based model fitting and prior specification are provided. An example, based on two small subsets of time intervals between eruptions of the Old Faithful geyser, illustrates the methodology.     

A stochastic interactive model for the diffusion of information

Two important generalizations of information diffusion models are the presence of stochastic effects and the possibility of arbitrary patterns of influence among individuals. A Markov random fields model includes both of these features. Under very weak assumptions, there is a unique equilibrium distribution of information patterns for given stochastic (local) interactions among a finite population. This has implications for policies to influence the transmission of information. The dynamic behavior of a special and simple case of the model tends to approximate the standard (logistic) diffusion curve. For an infinite population, uniqueness of equilibrium distributions may fail; some sufficient conditions to ensure uniqueness are given.     

Spatial defect pattern recognition on semiconductor wafers using model-based clustering and Bayesian inference

Defects on semiconductor wafers tend to cluster and the spatial defect patterns contain useful information about potential problems in the manufacturing process. This study proposes to use model-based clustering algorithms via Bayesian inferences for spatial defect pattern recognition on semiconductor wafers. These new algorithms can find the number of defect clusters as well as identify the pattern of each cluster automatically. They are capable of detecting curvilinear patterns, ellipsoidal patterns and nonuniform global defect patterns. Promising results have been obtained from simulation studies.     

Bayesian nonparametric modeling for functional analysis of variance

Analysis of variance is a standard statistical modeling approach for comparing populations. The functional analysis setting envisions that mean functions are associated with the populations, customarily modeled using basis representations, and seeks to compare them. Here, we adopt the modeling approach of functions as realizations of stochastic processes. We extend the Gaussian process version to allow nonparametric specifications using Dirichlet process mixing. Several metrics are introduced for comparison of populations. Then we introduce a hierarchical Dirichlet process model which enables comparison of the population distributions, either directly or through functionals of interest using the foregoing metrics. The modeling is extended to allow us to switch the sampling scheme. There are still population level distributions but now we sample at levels of the functions, obtaining observations from potentially different individuals at different levels. We illustrate with both simulated data and a dataset of temperature versus depth measurements at different locations in the Atlantic Ocean.     

Consensus in the two-state Axelrod model

The Axelrod model is a spatial stochastic model for the dynamics of cultures which, similar to the voter model, includes social influence, but differs from the latter by also accounting for another social factor called homophily, the tendency to interact more frequently with individuals who are more similar. Each individual is characterized by its opinions about a finite number of cultural features, each of which can assume the same finite number of states. Pairs of adjacent individuals interact at a rate equal to the fraction of features they have in common, thus modeling homophily, which results in the interacting pair having one more cultural feature in common, thus modeling social influence. It has been conjectured based on numerical simulations that the one-dimensional Axelrod model clusters when the number of features exceeds the number of states per feature. In this article, we prove this conjecture for the two-state model with an arbitrary number of features.     

Baseline models of spatial population dynamics

To understand human population dynamics fully, before considering complex human agency it may be useful to construct baseline models to see where such agency may and may not be necessary. In fact, the dynamics of human populations may be amenable to mathematical modeling with relatively parsimonious mechanisms. We review some of the more prominent of such models, namely, the spatial Galton-Watson (GW) model, modifications of the GW model that add migration and immigration, and the Bolker-Pacala model, in which mortality (or birth rate) is affected by competition. We show that change in the distribution of population density over the last century for 12 American rural states may be captured by the simplest of the models, the spatial GW model.     

ALLEE EFFECTS: POPULATION GROWTH,CRITICAL DENSITY,AND THE CHANCE OF EXTINCTION

This paper develops mathematical models to describe the growth, critical density, and extinction probability in sparse populations experiencing Allee effects. An Allee effect (or depensation) is a situation at low population densities where the per-individual growth rate is an increasing function of population density. A potentially important mechanism causing Allee effects is a shortage of mating encounters in sparse populations. Stochastic models are proposed for predicting the probability of encounter or the frequency of encounter as a function of population density. A negative exponential function is derived as such an encounter function under very general biological assumptions, including random, regular, or aggregated spatial patterns. A rectangular hyperbola function, heretofore used in ecology as the functional response of predator feeding rate to prey density, arises from the negative exponential function when encounter probabilities are assumed heterogeneous among individuals. These encounter functions produce Allee effects when incorporated into population growth models as birth rates. Three types of population models with encounter-limited birth rates are compared: (1) deterministic differential equations, (2) stochastic discrete birth-death processes, and (3) stochastic continuous diffusion processes. The phenomenon of a critical density, a major consequence of Allee effects, manifests itself differently in the different types of models. The critical density is a lower unstable equilibrium in the deterministic differential equation models. For the stochastic discrete birth-death processes considered here, the critical density is an inflection point in the probability of extinction plotted as a function of initial population density. In the continuous diffusion processes, the critical density becomes a local minimum (antimode) in the stationary probability distribution for population density. For both types of stochastic models, a critical density appears as an inflection point in the probability of attaining a small population density (extinction) before attaining a large one. Multiplicative (“environmental”) stochastic noise amplifies Allee effects. Harvesting also amplifies those effects. Though Allee effects are difficult to detect or measure in natural populations, their presence would seriously impact exploitation, management, and preservation of biological resources.     

Periodic autoregressive stochastic volatility

This paper proposes a stochastic volatility model (PAR-SV) in which the log-volatility follows a first-order periodic autoregression. This model aims at representing time series with volatility displaying a stochastic periodic dynamic structure, and may then be seen as an alternative to the familiar periodic GARCH process. The probabilistic structure of the proposed PAR-SV model such as periodic stationarity and autocovariance structure are first studied. Then, parameter estimation is examined through the quasi-maximum likelihood (QML) method where the likelihood is evaluated using the prediction error decomposition approach and Kalman filtering. In addition, a Bayesian MCMC method is also considered, where the posteriors are given from conjugate priors using the Gibbs sampler in which the augmented volatilities are sampled from the Griddy Gibbs technique in a single-move way. As a-by-product, period selection for the PAR-SV is carried out using the (conditional) deviance information criterion (DIC). A simulation study is undertaken to assess the performances of the QML and Bayesian Griddy Gibbs estimates in finite samples while applications of Bayesian PAR-SV modeling to daily, quarterly and monthly S&P 500 returns are considered.     

Bayesian reference analysis for Gaussian Markov random fields

Gaussian Markov random fields (GMRF) are important families of distributions for the modeling of spatial data and have been extensively used in different areas of spatial statistics such as disease mapping, image analysis and remote sensing. GMRFs have been used for the modeling of spatial data, both as models for the sampling distribution of the observed data and as models for the prior of latent processes/random effects; we consider mainly the former use of GMRFs. We study a large class of GMRF models that includes several models previously proposed in the literature. An objective Bayesian analysis is presented for the parameters of the above class of GMRFs, where explicit expressions for the Jeffreys (two versions) and reference priors are derived, and for each of these priors results on posterior propriety of the model parameters are established. We describe a simple MCMC algorithm for sampling from the posterior distribution of the model parameters, and study frequentist properties of the Bayesian inferences resulting from the use of these automatic priors. Finally, we illustrate the use of the proposed GMRF model and reference prior for studying the spatial variability of lip cancer cases in the districts of Scotland over the period 1975-1980.     

Modeling dynamics of cell population molecule expression distribution

This article introduces a novel approach to the study of the dynamics of the molecule expression level of large-size cell populations, whose goal is to understand how individual cell behavior propagates to population dynamics. A hybrid automaton framework is used which allows the simultaneous modeling of the formation and dissociation of cell-to-cell conjugations, and the molecular processes they control. Serial encounters among the cells are described by a stochastic approach under which the cell distribution over the state space is modeled and the dynamics of the state probability density functions is determined. This work is motivated by the investigation of T-cell receptor expression distribution. These receptors are essential for the antigen recognition and the regulation of the immune system. The results are illustrated with examples and validated with real data.     

Regression Tree Analysis Using TARGET

Regression trees are a popular alternative to classical regression methods. A number of approaches exist for constructing regression trees. Most of these techniques, including CART, are sequential in nature and locally optimal at each node split, so the final tree solution found may not be the best tree overall. In addition, small changes in the training data often lead to large changes in the final result due to the relative instability of these greedy tree-growing algorithms. Ensemble techniques, such as random forests, attempt to take advantage of this instability by growing a forest of trees from the data and averaging their predictions. The predictive performance is improved, but the simplicity of a single-tree solution is lost.

In earlier work, we introduced the Tree Analysis with Randomly Generated and Evolved Trees (TARGET) method for constructing classification trees via genetic algorithms. In this article, we extend the TARGET approach to regression trees. Simulated data and real world data are used to illustrate the TARGET process and compare its performance to CART, Bayesian CART, and random forests. The empirical results indicate that TARGET regression trees have better predictive performance than recursive partitioning methods, such as CART, and single-tree stochastic search methods, such as Bayesian CART. The predictive performance of TARGET is slightly worse than that of ensemble methods, such as random forests, but the TARGET solutions are far more interpretable.     

A Bayesian network model for spatial event surveillance

Methods for spatial cluster detection attempt to locate spatial subregions of some larger region where the count of some occurrences is higher than expected. Event surveillance consists of monitoring a region in order to detect emerging patterns that are indicative of some event of interest. In spatial event surveillance, we search for emerging patterns in spatial subregions.A well-known method for spatial cluster detection is Kulldorff’s [M. Kulldorff, A spatial scan statistic, Communications in Statistics: Theory and Methods 26 (6) (1997)] spatial scan statistic, which directly analyzes the counts of occurrences in the subregions. Neill et al. [D.B. Neill, A.W. Moore, G.F. Cooper, A Bayesian spatial scan statistic, Advances in Neural Information Processing Systems (NIPS) 18 (2005)] developed a Bayesian spatial scan statistic called BSS, which also directly analyzes the counts.We developed a new Bayesian-network-based spatial scan statistic, called BNetScan, which models the relationships among the events of interest and the observable events using a Bayesian network. BNetScan is an entity-based Bayesian network that models the underlying state and observable variables for each individual in a population.We compared the performance of BNetScan to Kulldorff’s spatial scan statistic and BSS using simulated outbreaks of influenza and cryptosporidiosis injected into real Emergency Department data from Allegheny County, Pennsylvania. It is an open question whether we can obtain acceptable results using a Bayesian network if the probability distributions in the network do not closely reflect reality, and thus, we examined the robustness of BNetScan relative to the probability distributions used to generate the data in the experiments. Our results indicate that BNetScan outperforms the other methods and its performance is robust relative to the probability distribution that is used to generate the data.     

NUMERICAL STUDY OF THE TWO-DIMENSIONAL SPRUCE BUDWORM REACTION-DIFFUSION EQUATION WITH DENSITY DEPENDENT DIFFUSION

The spruce budworm model is one of the interesting single species reaction-diffusion problems describing insect dispersal behavior. In this paper, we investigate a two-dimensional model with linear diffusion dependence and a convective wind. This system has been successfully solved using an operator splitting method for various domains and initial conditions. The numerical results show that populations can grow and diffuse in such a way as to produce steady state outbreak populations or steady state inhomogeneous spatial patterns in which they aggregate with low population densities.     

MODELING AND THE MANAGEMENT OF MIGRATORY BIRDS

Mathematical modeling of migratory bird populations is reviewed in the context of migratory bird management. We focus on dynamic models of waterfowl, since most management-oriented migratory bird models concern waterfowl species. We describe the management context for these modeling efforts, with a focus on large-scale operational data collection programs and on processes by which waterfowl harvest is regulated and waterfowl habitats are protected and managed. Through their impacts on key population parameters such as recruitment and survival rate, these activities can influence population dynamics, thereby providing managers some measure of control over the status of populations. Recent applications of the modeling of waterfowl are described in terms of objectives, mathematical structures, and contributions to management. Finally, we discuss research needs and data limitations in migratory bird modeling, and offer suggestions to increase the value to managers of future modeling efforts.     

On exploring the genetic algorithm for modeling the evolution of cooperation in a population

In this paper, we propose a genetic algorithm approximation for modeling a population which individuals compete with each other based on prisoner’s dilemma game. Players act according to their genome, which gives them a strategy (phenotype); in our case, a probability for cooperation. The most successful players will produce more offspring and that depends directly of the strategy adopted. As individuals die, the newborns parents will be those fittest individuals in a certain spatial region. Four different fitness functions are tested to investigate the influence in the evolution of cooperation. Individuals live in a lattice modeled by probabilistic cellular automata and play the game with some of their neighborhoods. In spite of players homogeneously distributed over the space, a mean-field approximation is presented in terms of ordinary differential equations.     

Approximating the ideal free distribution via reaction-diffusion-advection equations

We consider reaction-diffusion-advection models for spatially distributed populations that have a tendency to disperse up the gradient of fitness, where fitness is defined as a logistic local population growth rate. We show that in temporally constant but spatially varying environments such populations have equilibrium distributions that can approximate those that would be predicted by a version of the ideal free distribution incorporating population dynamics. The modeling approach shows that a dispersal mechanism based on local information about the environment and population density can approximate the ideal free distribution. The analysis suggests that such a dispersal mechanism may sometimes be advantageous because it allows populations to approximately track resource availability. The models are quasilinear parabolic equations with nonlinear boundary conditions.     

Empirical Likelihood Estimation for the Two-Sample Mean under Density Ratio Models Using Auxiliary Information

Auxiliary population information is often available in finite population inference problems, and the empirical likelihood (EL) approach has been demonstrated to be flexible and useful for such problems. The present paper concerns EL when interest centers on inference for the mean of the baseline distribution under two-sample density ratio models. Although dual EL is a convenient technical tool since it has the same maximum point and maximum likelihood as DRM-based EL, it can not combine such auxiliary information into the likelihood conveniently and may have loss of efficiency. By contrast, the classical EL approach of Qin and Lawless\ucite{21} does not have this problem and incorporate seamlessly auxiliary information. Based on the EL using auxiliary information and the dual EL methods, we construct both point and interval estimations and make a careful comparison. Though the point estimation efficiency gain obtained by the former is not noticeable, we find that they may have different performances in interval estimation. In terms of coverage accuracy, the two intervals are comparable for not or moderate skewed populations, and the EL interval using auxiliary information can be much superior for severely skewed populations.