The analysis of categorical data often leads to the analysis of a contingency table. For large samples, asymptotic approximations are sufficient when calculating p-values, but for small samples the tests can be unreliable. In these situations an exact test should be considered. This bases the test on the exact distribution of the test statistic. Sampling techniques can be used to estimate the distribution. Alternatively, the distribution can be found by complete enumeration. A new algorithm is developed that enables a model to be defined by a model matrix, and all tables that satisfy the model are found. This provides a more efficient enumeration mechanism for complex models and extends the range of models that can be tested. The technique can lead to large calculations and a distributed version of the algorithm is developed that enables a number of machines to work efficiently on the same problem.

The purpose of this paper is to compare Analysis of Variance with Contingency Table Analysis when the data being analyzed do not satisfy Analysis of Variance assumptions. The criteria for comparison are the powers of the Standard variance-ratio and the Chi-square test. The test statistic and powers were obtained by Monte Carlo. 1. Calculate test statistic for each of 100 trials, this process was repeated 12 times. Each time different combination of means and variances were used. 2. Powers were obtained for each of 12 combinations of means and variances. Whether Analysis of Variance or Contingency Table Analysis is a better alternative depends on if we are interested in equality of population means or differences of population variances.

The aim of the diploma thesis was to create independently an application, automating a correspondence analysis calculation using MS Excel and taking advantage of VBA programming language. The application was then used to analyse Slovak banking sector. The created application is based on macro sets which can be split up to two parts. The first part produces a contingency table and converts it into a format usable in the second part. Then there is executed the Correspondence Analysis calculation itself. Supplement Matrix is being used, that allows to discharge functions from matrix and linear algebra. The application allows to process any matrix dimension. Dialog window offers four normalization types, shortened and also full output and it is possible to display row or column variable categories, eventually both. The analysis subject was to find out which of the products are typical for considered banks and for which clients, divided into age categories, is attractive particular bank and/or product. A starting point is a product usage data matrix at a particular bank. There was made a conclusion, that it is possible to divide banks into more traditional or modern type institutions and also that each bank has a typical product.

<p>Many modern applications fall into the category of "large-scale" statistical problems, in which both the number of observations n and the number of features or parameters p may be large. Many existing methods focus on point estimation, despite the continued relevance of uncertainty quantification in the sciences, where the number of parameters to estimate often exceeds the sample size, despite huge increases in the value of n typically seen in many fields. Thus, the tendency in some areas of industry to dispense with traditional statistical analysis on the basis that "n=all" is of little relevance outside of certain narrow applications. The main result of the Big Data revolution in most fields has instead been to make computation much harder without reducing the importance of uncertainty quantification. Bayesian methods excel at uncertainty quantification, but often scale poorly relative to alternatives. This conflict between the statistical advantages of Bayesian procedures and their substantial computational disadvantages is perhaps the greatest challenge facing modern Bayesian statistics, and is the primary motivation for the work presented here. </p><p>Two general strategies for scaling Bayesian inference are considered. The first is the development of methods that lend themselves to faster computation, and the second is design and characterization of computational algorithms that scale better in n or p. In the first instance, the focus is on joint inference outside of the standard problem of multivariate continuous data that has been a major focus of previous theoretical work in this area. In the second area, we pursue strategies for improving the speed of Markov chain Monte Carlo algorithms, and characterizing their performance in large-scale settings. Throughout, the focus is on rigorous theoretical evaluation combined with empirical demonstrations of performance and concordance with the theory.</p><p>One topic we consider is modeling the joint distribution of multivariate categorical data, often summarized in a contingency table. Contingency table analysis routinely relies on log-linear models, with latent structure analysis providing a common alternative. Latent structure models lead to a reduced rank tensor factorization of the probability mass function for multivariate categorical data, while log-linear models achieve dimensionality reduction through sparsity. Little is known about the relationship between these notions of dimensionality reduction in the two paradigms. In Chapter 2, we derive several results relating the support of a log-linear model to nonnegative ranks of the associated probability tensor. Motivated by these findings, we propose a new collapsed Tucker class of tensor decompositions, which bridge existing PARAFAC and Tucker decompositions, providing a more flexible framework for parsimoniously characterizing multivariate categorical data. Taking a Bayesian approach to inference, we illustrate empirical advantages of the new decompositions.</p><p>Latent class models for the joint distribution of multivariate categorical, such as the PARAFAC decomposition, data play an important role in the analysis of population structure. In this context, the number of latent classes is interpreted as the number of genetically distinct subpopulations of an organism, an important factor in the analysis of evolutionary processes and conservation status. Existing methods focus on point estimates of the number of subpopulations, and lack robust uncertainty quantification. Moreover, whether the number of latent classes in these models is even an identified parameter is an open question. In Chapter 3, we show that when the model is properly specified, the correct number of subpopulations can be recovered almost surely. We then propose an alternative method for estimating the number of latent subpopulations that provides good quantification of uncertainty, and provide a simple procedure for verifying that the proposed method is consistent for the number of subpopulations. The performance of the model in estimating the number of subpopulations and other common population structure inference problems is assessed in simulations and a real data application.</p><p>In contingency table analysis, sparse data is frequently encountered for even modest numbers of variables, resulting in non-existence of maximum likelihood estimates. A common solution is to obtain regularized estimates of the parameters of a log-linear model. Bayesian methods provide a coherent approach to regularization, but are often computationally intensive. Conjugate priors ease computational demands, but the conjugate Diaconis--Ylvisaker priors for the parameters of log-linear models do not give rise to closed form credible regions, complicating posterior inference. In Chapter 4 we derive the optimal Gaussian approximation to the posterior for log-linear models with Diaconis--Ylvisaker priors, and provide convergence rate and finite-sample bounds for the Kullback-Leibler divergence between the exact posterior and the optimal Gaussian approximation. We demonstrate empirically in simulations and a real data application that the approximation is highly accurate, even in relatively small samples. The proposed approximation provides a computationally scalable and principled approach to regularized estimation and approximate Bayesian inference for log-linear models. </p><p>Another challenging and somewhat non-standard joint modeling problem is inference on tail dependence in stochastic processes. In applications where extreme dependence is of interest, data are almost always time-indexed. Existing methods for inference and modeling in this setting often cluster extreme events or choose window sizes with the goal of preserving temporal information. In Chapter 5, we propose an alternative paradigm for inference on tail dependence in stochastic processes with arbitrary temporal dependence structure in the extremes, based on the idea that the information on strength of tail dependence and the temporal structure in this dependence are both encoded in waiting times between exceedances of high thresholds. We construct a class of time-indexed stochastic processes with tail dependence obtained by endowing the support points in de Haan's spectral representation of max-stable processes with velocities and lifetimes. We extend Smith's model to these max-stable velocity processes and obtain the distribution of waiting times between extreme events at multiple locations. Motivated by this result, a new definition of tail dependence is proposed that is a function of the distribution of waiting times between threshold exceedances, and an inferential framework is constructed for estimating the strength of extremal dependence and quantifying uncertainty in this paradigm. The method is applied to climatological, financial, and electrophysiology data. </p><p>The remainder of this thesis focuses on posterior computation by Markov chain Monte Carlo. The Markov Chain Monte Carlo method is the dominant paradigm for posterior computation in Bayesian analysis. It has long been common to control computation time by making approximations to the Markov transition kernel. Comparatively little attention has been paid to convergence and estimation error in these approximating Markov Chains. In Chapter 6, we propose a framework for assessing when to use approximations in MCMC algorithms, and how much error in the transition kernel should be tolerated to obtain optimal estimation performance with respect to a specified loss function and computational budget. The results require only ergodicity of the exact kernel and control of the kernel approximation accuracy. The theoretical framework is applied to approximations based on random subsets of data, low-rank approximations of Gaussian processes, and a novel approximating Markov chain for discrete mixture models.</p><p>Data augmentation Gibbs samplers are arguably the most popular class of algorithm for approximately sampling from the posterior distribution for the parameters of generalized linear models. The truncated Normal and Polya-Gamma data augmentation samplers are standard examples for probit and logit links, respectively. Motivated by an important problem in quantitative advertising, in Chapter 7 we consider the application of these algorithms to modeling rare events. We show that when the sample size is large but the observed number of successes is small, these data augmentation samplers mix very slowly, with a spectral gap that converges to zero at a rate at least proportional to the reciprocal of the square root of the sample size up to a log factor. In simulation studies, moderate sample sizes result in high autocorrelations and small effective sample sizes. Similar empirical results are observed for related data augmentation samplers for multinomial logit and probit models. When applied to a real quantitative advertising dataset, the data augmentation samplers mix very poorly. Conversely, Hamiltonian Monte Carlo and a type of independence chain Metropolis algorithm show good mixing on the same dataset.</p> / Dissertation

This thesis consists of four papers that deal with several aspects of the measurement of model fit for categorical data. In all papers, special attention is paid to situations with sparse data. The first paper concerns the computational burden of calculating Pearson's goodness-of-fit statistic for situations where many response patterns have observed frequencies that equal zero. A simple solution is presented that allows for the computation of the total value of Pearson's goodness-of-fit statistic when the expected frequencies of response patterns with observed frequencies of zero are unknown. In the second paper, a new fit statistic is presented that is a modification of Pearson's statistic but that is not adversely affected by response patterns with very small expected frequencies. It is shown that the new statistic is asymptotically equivalent to Pearson's goodness-of-fit statistic and hence, asymptotically chi-square distributed. In the third paper, comprehensive simulation studies are conducted that compare seven asymptotically equivalent fit statistics, including the new statistic. Situations that are considered concern both multinomial sampling and factor analysis. Tests for the goodness-of-fit are conducted by means of the asymptotic and the bootstrap approach both under the null hypothesis and when there is a certain degree of misfit in the data. Results indicate that recommendations on the use of a fit statistic can be dependent on the investigated situation and on the purpose of the model test. Power varies substantially between the fit statistics and the cause of the misfit of the model. Findings indicate further that the new statistic proposed in this thesis shows rather stable results and compared to the other fit statistics, no disadvantageous characteristics of the fit statistic are found. Finally, in the fourth paper, the potential necessity of determining the goodness-of-fit by two sided model testing is adverted. A simulation study is conducted that investigates differences between the one sided and the two sided approach of model testing. Situations are identified for which two sided model testing has advantages over the one sided approach.

One of the oldest and most fundamental problems in statistics is the analysis of cross-classified data called contingency tables. Analyzing contingency tables is typically a question of association - do the variables represented in the table exhibit special dependencies or lack thereof? The statistical models which best capture these experimental notions of dependence are the categorical conditional independence models; however, until recent discoveries concerning the strongly algebraic nature of the conditional independence models surfaced, the models were widely overlooked due to their unwieldy implicit description. Apart from the inferential question above, this thesis asks the more basic question - suppose such an experimental model of association is known, how can one incorporate this information into the estimation of the joint distribution of the table? In the traditional parametric setting several estimation paradigms have been developed over the past century; however, traditional results are not applicable to arbitrary categorical conditional independence models due to their implicit nature. After laying out the framework for conditional independence and algebraic statistical models, we consider three aspects of estimation in the models using the minimum Euclidean (L2E), minimum Pearson chi-squared, and minimum Neyman modified chi-squared distance paradigms as well as the more ubiquitous maximum likelihood approach (MLE). First, we consider the theoretical properties of the estimators and demonstrate that under general conditions the estimators exist and are asymptotically normal. For small samples, we present the results of large scale simulations to address the estimators' bias and mean squared error (in the Euclidean and Frobenius norms, respectively). Second, we identify the computation of such estimators as an optimization problem and, for the case of the L2E, propose two different methods by which the problem can be solved, one algebraic and one numerical. Finally, we present an R implementation via two novel packages, mpoly for symbolic computing with multivariate polynomials and catcim for fitting categorical conditional independence models. It is found that in general minimum distance estimators in categorical conditional independence models behave as they do in the more traditional parametric setting and can be computed in many practical situations with the implementation provided.

<p>Identifying a lower-dimensional latent space for representation of high-dimensional observations is of significant importance in numerous biomedical and machine learning applications. In many such applications, it is now routine to collect data where the dimensionality of the outcomes is comparable or even larger than the number of available observations. Motivated in particular by the problem of predicting the risk of impending diseases from massive gene expression and single nucleotide polymorphism profiles, this dissertation focuses on building parsimonious models and computational schemes for high-dimensional continuous and unordered categorical data, while also studying theoretical properties of the proposed methods. Sparse factor modeling is fast becoming a standard tool for parsimonious modeling of such massive dimensional data and the content of this thesis is specifically directed towards methodological and theoretical developments in Bayesian sparse factor models.</p><p>The first three chapters of the thesis studies sparse factor models for high-dimensional continuous data. A class of shrinkage priors on factor loadings are introduced with attractive computational properties, with operating characteristics explored through a number of simulated and real data examples. In spite of the methodological advances over the past decade, theoretical justifications in high-dimensional factor models are scarce in the Bayesian literature. Part of the dissertation focuses on exploring estimation of high-dimensional covariance matrices using a factor model and studying the rate of posterior contraction as both the sample size & dimensionality increases. </p><p>To relax the usual assumption of a linear relationship among the latent and observed variables in a standard factor model, extensions to a non-linear latent factor model are also considered.</p><p>Although Gaussian latent factor models are routinely used for modeling of dependence in continuous, binary and ordered categorical data, it leads to challenging computation and complex modeling structures for unordered categorical variables. As an alternative, a novel class of simplex factor models for massive-dimensional and enormously sparse contingency table data is proposed in the second part of the thesis. An efficient MCMC scheme is developed for posterior computation and the methods are applied to modeling dependence in nucleotide sequences and prediction from high-dimensional categorical features. Building on a connection between the proposed model & sparse tensor decompositions, we propose new classes of nonparametric Bayesian models for testing associations between a massive dimensional vector of genetic markers and a phenotypical outcome.</p> / Dissertation

京都大学 / 新制・課程博士 / 博士(医学) / 甲第23367号 / 医博第4736号 / 新制||医||1051(附属図書館) / 京都大学大学院医学研究科医学専攻 / (主査)教授 森田 智視, 教授 川上 浩司, 教授 佐藤 俊哉 / 学位規則第4条第1項該当 / Doctor of Medical Science / Kyoto University / DFAM

Odds ratios and risk ratios are useful measures of effect size in 2-group studies in which the response variable is dichotomous. Confidence interval methods are proposed for combining and comparing odds ratios and risk ratios in multistudy designs. Unlike the traditional fixed-effect meta-analysis methods, the proposed varying coefficient methods do not require effect-size homogeneity, and unlike the randomeffects meta-analysis methods, the proposed varying coefficient methods do not assume that the effect sizes from the selected studies represent a random sample from a normally distributed superpopulation of effect sizes. The results of extensive simulation studies suggest that the proposed varying coefficient methods have excellent performance characteristics under realistic conditions and should provide useful alternatives to the currently used meta-analysis methods.

Point estimation of several association parameters in categorical data are presented. Typically, a constant is added to the frequency counts before the association measure is computed. We will study the accuracy of these adjusted point estimators based on frequentist and Bayesian methods respectively. In particular, amended estimators for the ratio of independent Poisson rates, relative risk, odds ratio, and the ratio of marginal binomial proportions will be examined in terms of bias and mean squared error.