Model Similarity and Rank-Order Based Classification of Bayesian Networks

Suppose that we rank-order the conditional probabilities for a group of subjects that are provided from a Bayesian network (BN) model of binary variables. The conditional probability is the probability that a subject has a certain attribute given an outcome of some other variables and the classification is based on the rank-order. Under the condition that the class sizes are equal across the class levels and that all the variables in the model are positively associated with each other, we compared the classification results between models of binary variables which share the same model structure. In the comparison, we used a BN model, called a similar BN model, which was constructed under some rule based on a set of BN models satisfying certain conditions. Simulation results indicate that the agreement level of the classification between a set of BN models and their corresponding similar BN model is considerably high with the exact agreement for about half of the subjects or more and the agreement up to one-class-level difference for about 90% or more.     

The location model for mixtures of categorical and continuous variables

Recent research into graphical association models has focussed interest on the conditional Gaussian distribution for analyzing mixtures of categorical and continuous variables. A special case of such models, utilizing the homogeneous conditional Gaussian distribution, has in fact been known since 1961 as the location model, and for the past 30 years has provided a basis for the multivariate analysis of mixed categorical and continuous variables. Extensive development of this model took place throughout the 1970’s and 1980’s in the context of discrimination and classification, and comprehensive methodology is now available for such analysis of mixed variables. This paper surveys these developments and summarizes current capabilities in the area. Topics include distances between groups, discriminant analysis, error rates and their estimation, model and feature selection, and the handling of missing data.     

Dimensionally Reduced Model-Based Clustering Through Mixtures of Factor Mixture Analyzers

Dimensionally reduced model-based clustering methods are recently receiving a wide interest in statistics as a tool for performing simultaneously clustering and dimension reduction through one or more latent variables. Among these, Mixtures of Factor Analyzers assume that, within each component, the data are generated according to a factor model, thus reducing the number of parameters on which the covariance matrices depend. In Factor Mixture Analysis clustering is performed through the factors of an ordinary factor analysis which are jointly modelled by a Gaussian mixture. The two approaches differ in genesis, parameterization and consequently clustering performance. In this work we propose a model which extends and combines them. The proposed Mixtures of Factor Mixture Analyzers provide a unified class of dimensionally reduced mixture models which includes the previous ones as special cases and could offer a powerful tool for modelling non-Gaussian latent variables.     

Divisive Latent Class Modeling as a Density Estimation Method for Categorical Data

Traditionally latent class (LC) analysis is used by applied researchers as a tool for identifying substantively meaningful clusters. More recently, LC models have also been used as a density estimation tool for categorical variables. We introduce a divisive LC (DLC) model as a density estimation tool that may offer several advantages in comparison to a standard LC model. When using an LC model for density estimation, a considerable number of increasingly large LC models may have to be estimated before sufficient model-fit is achieved. A DLC model consists of a sequence of small LC models. Therefore, a DLC model can be estimated much faster and can easily utilize multiple processor cores, meaning that this model is more widely applicable and practical. In this study we describe the algorithm of fitting a DLC model, and discuss the various settings that indirectly influence the precision of a DLC model as a density estimation tool. These settings are illustrated using a synthetic data example, and the best performing algorithm is applied to a real-data example. The generated data example showed that, using specific decision rules, a DLC model is able to correctly model complex associations amongst categorical variables.     

Enhanced Model-Based Clustering,Density Estimation,and Discriminant Analysis Software: MCLUST

MCLUST is a software package for model-based clustering, density estimation and discriminant analysis interfaced to the S-PLUS commercial software and the R language. It implements parameterized Gaussian hierarchical clustering algorithms and the EM algorithm for parameterized Gaussian mixture models with the possible addition of a Poisson noise term. Also included are functions that combine hierarchical clustering, EM and the Bayesian Information Criterion (BIC) in comprehensive strategies for clustering, density estimation, and discriminant analysis. MCLUST provides functionality for displaying and visualizing clustering and classification results. A web page with related links can be found at .     

Measuring the Reliability of Diagnostic Classification Model Examinee Estimates

Over the past decade, diagnostic classification models (DCMs) have become an active area of psychometric research. Despite their use, the reliability of examinee estimates in DCM applications has seldom been reported. In this paper, a reliability measure for the categorical latent variables of DCMs is defined. Using theory-and simulation-based results, we show how DCMs uniformly provide greater examinee estimate reliability than IRT models for tests of the same length, a result that is a consequence of the smaller range of latent variable values examinee estimates can take in DCMs. We demonstrate this result by comparing DCM and IRT reliability for a series of models estimated with data from an end-of-grade test, culminating with a discussion of how DCMs can be used to change the character of large scale testing, either by shortening tests that measure examinees unidimensionally or by providing more reliable multidimensional measurement for tests of the same length.     

Simultaneous Predictive Gaussian Classifiers

Gaussian distribution has for several decades been ubiquitous in the theory and practice of statistical classification. Despite the early proposals motivating the use of predictive inference to design a classifier, this approach has gained relatively little attention apart from certain specific applications, such as speech recognition where its optimality has been widely acknowledged. Here we examine statistical properties of different inductive classification rules under a generic Gaussian model and demonstrate the optimality of considering simultaneous classification of multiple samples under an attractive loss function. It is shown that the simpler independent classification of samples leads asymptotically to the same optimal rule as the simultaneous classifier when the amount of training data increases, if the dimensionality of the feature space is bounded in an appropriate manner. Numerical investigations suggest that the simultaneous predictive classifier can lead to higher classification accuracy than the independent rule in the low-dimensional case, whereas the simultaneous approach suffers more from noise when the dimensionality increases.     

Incorporating Student Covariates in Cognitive Diagnosis Models

In educational measurement, cognitive diagnosis models have been developed to allow assessment of specific skills that are needed to perform tasks. Skill knowledge is characterized as present or absent and represented by a vector of binary indicators, or the skill set profile. After determining which skills are needed for each assessment item, a model is specified for the relationship between item responses and skill set profiles. Cognitive diagnosis models are often used for diagnosis, that is, for classifying students into the different skill set profiles. Generally, cognitive diagnosis models do not exploit student covariate information. However, investigating the effects of student covariates, such as gender, SES, or educational interventions, on skill knowledge mastery is important in education research, and covariate information may improve classification of students to skill set profiles. We extend a common cognitive diagnosis model, the DINA model, by modeling the relationship between the latent skill knowledge indicators and covariates. The probability of skill mastery is modeled as a logistic regression model, possibly with a student-level random intercept, giving a higher-order DINA model with a latent regression. Simulations show that parameter recovery is good for these models and that inclusion of covariates can improve skill diagnosis. When applying our methods to data from an online tutor, we obtain reasonable and interpretable parameter estimates that allow more detailed characterization of groups of students who differ in their predicted skill set profiles.     

A Property of the CHAID Partitioning Method for Dichotomous Randomized Response Data and Categorical Predictors

In this paper, we present empirical and theoretical results on classification trees for randomized response data. We considered a dichotomous sensitive response variable with the true status intentionally misclassified by the respondents using rules prescribed by a randomized response method. We assumed that classification trees are grown using the Pearson chi-square test as a splitting criterion, and that the randomized response data are analyzed using classification trees as if they were not perturbed. We proved that classification trees analyzing observed randomized response data and estimated true data have a one-to-one correspondence in terms of ranking the splitting variables. This is illustrated using two real data sets.     

Optimal Variable Weighting for Ultrametric and Additive Trees and K-means Partitioning: Methods and Software

K -means partitioning. We also describe some new features and improvements to the algorithm proposed by De Soete. Monte Carlo simulations have been conducted using different error conditions. In all cases (i.e., ultrametric or additive trees, or K-means partitioning), the simulation results indicate that the optimal weighting procedure should be used for analyzing data containing noisy variables that do not contribute relevant information to the classification structure. However, if the data involve error-perturbed variables that are relevant to the classification or outliers, it seems better to cluster or partition the entities by using variables with equal weights. A new computer program, OVW, which is available to researchers as freeware, implements improved algorithms for optimal variable weighting for ultrametric and additive tree clustering, and includes a new algorithm for optimal variable weighting for K-means partitioning.     

Confidence in Classification: A Bayesian Approach

Bayesian classification is currently of considerable interest. It provides a strategy for eliminating the uncertainty associated with a particular choice of classifiermodel parameters, and is the optimal decision-theoretic choice under certain circumstances when there is no single “true” classifier for a given data set. Modern computing capabilities can easily support the Markov chain Monte Carlo sampling that is necessary to carry out the calculations involved, but the information available in these samples is not at present being fully utilised. We show how it can be allied to known results concerning the “reject option” in order to produce an assessment of the confidence that can be ascribed to particular classifications, and how these confidence measures can be used to compare the performances of classifiers. Incorporating these confidence measures can alter the apparent ranking of classifiers as given by straightforward success or error rates. Several possible methods for obtaining confidence assessments are described, and compared on a range of data sets using the Bayesian probabilistic nearest-neighbour classifier.     

Maximum Likelihood Estimation and Model Comparison for Mixtures of Structural Equation Models with Ignorable Missing Data

The objective of this paper is to develop the maximum likelihood approach for analyzing a finite mixture of structural equation models with missing data that are missing at random. A Monte Carlo EM algorithm is proposed for obtaining the maximum likelihood estimates. A well-known statistic in model comparison, namely the Bayesian Information Criterion (BIC), is used for model comparison. With the presence of missing data, the computation of the observed-data likelihood function value involved in the BIC is not straightforward. A procedure based on path sampling is developed to compute this function value. It is shown by means of simulation studies that ignoring the incomplete data with missing entries gives less accurate ML estimates. An illustrative real example is also presented.     

Power and Sample Size Computation for Wald Tests in Latent Class Models

Latent class (LC) analysis is used by social, behavioral, and medical science researchers among others as a tool for clustering (or unsupervised classification) with categorical response variables, for analyzing the agreement between multiple raters, for evaluating the sensitivity and specificity of diagnostic tests in the absence of a gold standard, and for modeling heterogeneity in developmental trajectories. Despite the increased popularity of LC analysis, little is known about statistical power and required sample size in LC modeling. This paper shows how to perform power and sample size computations in LC models using Wald tests for the parameters describing association between the categorical latent variable and the response variables. Moreover, the design factors affecting the statistical power of these Wald tests are studied. More specifically, we show how design factors which are specific for LC analysis, such as the number of classes, the class proportions, and the number of response variables, affect the information matrix. The proposed power computation approach is illustrated using realistic scenarios for the design factors. A simulation study conducted to assess the performance of the proposed power analysis procedure shows that it performs well in all situations one may encounter in practice.     

A Probability for Classification Based on the Dirichlet Process Mixture Model

In this paper we provide an explicit probability distribution for classification purposes when observations are viewed on the real line and classifications are to be based on numerical orderings. The classification model is derived from a Bayesian nonparametric mixture of Dirichlet process model; with some modifications. The resulting approach then more closely resembles a classical hierarchical grouping rule in that it depends on sums of squares of neighboring values. The proposed probability model for classification relies on a numerical procedure based on a reversible Markov chain Monte Carlo (MCMC) algorithm for determining the probabilities. Some numerical illustrations comparing with alternative ideas for classification are provided.     

A Hierarchical Methodology for Class Detection Problems with Skewed Priors

We describe a novel extension to the Class-Cover-Catch-Digraph (CCCD) classifier, specifically tuned to detection problems. These are two-class classification problems where the natural priors on the classes are skewed by several orders of magnitude. The emphasis of the proposed techniques is in computationally efficient classification for real-time applications. Our principal contribution consists of two boosted classi- fiers built upon the CCCD structure, one in the form of a sequential decision process and the other in the form of a tree. Both of these classifiers achieve performances comparable to that of the original CCCD classifiers, but at drastically reduced computational expense. An analysis of classification performance and computational cost is performed using data from a face detection application. Comparisons are provided with Support Vector Machines (SVM) and reduced SVMs. These comparisons show that while some SVMs may achieve higher classification performance, their computational burden can be so high as to make them unusable in real-time applications. On the other hand, the proposed classifiers combine high detection performance with extremely fast classification.     

Bayes Discriminant Rules with Ordered Predictors

We propose and discuss improved Bayes rules to discriminate between two populations using ordered predictors. To address the problem we propose an alternative formulation using a latent space that allows to introduce the information about the order in the theoretical rules. The rules are first defined when the marginal densities are fully known and then under normality when the parameters are unknown and training samples are available. Several numerical examples and simulations in the paper illustrate the methodology and show that the new rules handle the information appropriately. We compare the new rules with the classical Bayes and Fisher rules in these examples and we show that the misclassification probability is smaller for the new rules. The method is also applied to data from a diabetes study where we again show that the new rules improve over the usual Fisher rule. Research partially supported by Spanish DGES and by PAPIJCL. The authors thank the editor and an anonymous reviewer for their detailed reading that resulted in this much improved version of the paper.     

The Academic Journal Ranking Problem: A Fuzzy-Clustering Approach

The Academic Journal Ranking Problem consists in formulating a formal assessment of scientific journals. An outcome variable must be constructed that allows valid journal comparison, either as a set of tiers (ordered classes) or as a numerical index. But part of the problem is also to devise a procedure to get this outcome, that is, how to get and use relevant data coming from expert opinions or from citations database. We propose a novel approach to the problem that applies fuzzy cluster analysis to peer reviews and opinion surveys. The procedure is composed of two steps: the first is to collect the most relevant qualitative assessments from international organizations (for example, the ones available in the Harzing database) and, as inductive analysis, to apply fuzzy clustering to determine homogeneous journal classes; the second deductive step is to determine the hidden logical rules that underlies the classification, using a classification tree to reproduce the same patterns of the first step.     

Model-Based Clustering

The notion of defining a cluster as a component in a mixture model was put forth by Tiedeman in 1955; since then, the use of mixture models for clustering has grown into an important subfield of classification. Considering the volume of work within this field over the past decade, which seems equal to all of that which went before, a review of work to date is timely. First, the definition of a cluster is discussed and some historical context for model-based clustering is provided. Then, starting with Gaussian mixtures, the evolution of model-based clustering is traced, from the famous paper by Wolfe in 1965 to work that is currently available only in preprint form. This review ends with a look ahead to the next decade or so.     

A comparison between two distance-based discriminant principles

A distance-based classification procedure suggested by Matusita (1956) has long been available as an alternative to the usual Bayes decision rule. Unsatisfactory features of both approaches when applied to multinomial data led Goldstein and Dillon (1978) to propose a new distance-based principle for classification. We subject the Goldstein/Dillon principle to some theoretical scrutiny by deriving the population classification rules appropriate not only to multinomial data but also to multivariate normal and mixed multinomial/multinormal data. These rules demonstrate equivalence of the Goldstein/Dillon and Matusita approaches for the first two data types, and similar equivalence is conjectured (but not explicitly obtained) for the mixed data case. Implications for sample-based rules are noted.     

Finite Mixture Modeling of Gaussian Regression Time Series with Application to Dendrochronology

Finite mixture modeling is a popular statistical technique capable of accounting for various shapes in data. One popular application of mixture models is model-based clustering. This paper considers the problem of clustering regression autoregressive moving average time series. Two novel estimation procedures for the considered framework are developed. The first one yields the conditional maximum likelihood estimates which can be used in cases when the length of times series is substantial. Simple analytical expressions make fast parameter estimation possible. The second method incorporates the Kalman filter and yields the exact maximum likelihood estimates. The procedure for assessing variability in obtained estimates is discussed. We also show that the Bayesian information criterion can be successfully used to choose the optimal number of mixture components and correctly assess time series orders. The performance of the developed methodology is evaluated on simulation studies. An application to the analysis of tree ring data is thoroughly considered. The results are very promising as the proposed approach overcomes the limitations of other methods developed so far.