Boosting with Bayesian stumps

Boosting is a new, powerful method for classification. It is an iterative procedure which successively classifies a weighted version of the sample, and then reweights this sample dependent on how successful the classification was. In this paper we review some of the commonly used methods for performing boosting and show how they can be fit into a Bayesian setup at each iteration of the algorithm. We demonstrate how this formulation gives rise to a new splitting criterion when using a domain-partitioning classification method such as a decision tree. Further we can improve the predictive performance of simple decision trees, known as stumps, by using a posterior weighted average of them to classify at each step of the algorithm, rather than just a single stump. The main advantage of this approach is to reduce the number of boosting iterations required to produce a good classifier with only a minimal increase in the computational complexity of the algorithm.     

Nonparametric empirical Bayes for the Dirichlet process mixture model

The Dirichlet process prior allows flexible nonparametric mixture modeling. The number of mixture components is not specified in advance and can grow as new data arrive. However, analyses based on the Dirichlet process prior are sensitive to the choice of the parameters, including an infinite-dimensional distributional parameter G ₀. Most previous applications have either fixed G ₀ as a member of a parametric family or treated G ₀ in a Bayesian fashion, using parametric prior specifications. In contrast, we have developed an adaptive nonparametric method for constructing smooth estimates of G ₀. We combine this method with a technique for estimating α, the other Dirichlet process parameter, that is inspired by an existing characterization of its maximum-likelihood estimator. Together, these estimation procedures yield a flexible empirical Bayes treatment of Dirichlet process mixtures. Such a treatment is useful in situations where smooth point estimates of G ₀ are of intrinsic interest, or where the structure of G ₀ cannot be conveniently modeled with the usual parametric prior families. Analysis of simulated and real-world datasets illustrates the robustness of this approach.     

Defining attributes for scorecard construction in credit scoring

In many domains, simple forms of classification rules are needed because of requirements such as ease of use. A particularly simple form splits each variable into just a few categories, assigns weights to the categories, sums the weights for a new object to be classified, and produces a classification by comparing the score with a threshold. Such instruments are often called scorecards. We describe a way to find the best partition of each variable using a simulated annealing strategy. We present theoretical and empirical comparisons of two such additive models, one based on weights of evidence and another based on logistic regression.     

A re-evaluation of random-effects meta-analysis

Summary.  Meta-analysis in the presence of unexplained heterogeneity is frequently undertaken by using a random-effects model, in which the effects underlying different studies are assumed to be drawn from a normal distribution. Here we discuss the justification and interpretation of such models, by addressing in turn the aims of estimation, prediction and hypothesis testing. A particular issue that we consider is the distinction between inference on the mean of the random-effects distribution and inference on the whole distribution. We suggest that random-effects meta-analyses as currently conducted often fail to provide the key results, and we investigate the extent to which distribution-free, classical and Bayesian approaches can provide satisfactory methods. We conclude that the Bayesian approach has the advantage of naturally allowing for full uncertainty, especially for prediction. However, it is not without problems, including computational intensity and sensitivity to a priori judgements. We propose a simple prediction interval for classical meta-analysis and offer extensions to standard practice of Bayesian meta-analysis, making use of an example of studies of 'set shifting' ability in people with eating disorders.     

Invariant co-ordinate selection

Summary.  A general method for exploring multivariate data by comparing different estimates of multivariate scatter is presented. The method is based on the eigenvalue–eigenvector decomposition of one scatter matrix relative to another. In particular, it is shown that the eigenvectors can be used to generate an affine invariant co-ordinate system for the multivariate data. Consequently, we view this method as a method for invariant co-ordinate selection . By plotting the data with respect to this new invariant co-ordinate system, various data structures can be revealed. For example, under certain independent components models, it is shown that the invariant co- ordinates correspond to the independent components. Another example pertains to mixtures of elliptical distributions. In this case, it is shown that a subset of the invariant co-ordinates corresponds to Fisher's linear discriminant subspace, even though the class identifications of the data points are unknown. Some illustrative examples are given.     

The Pacific Rim Library: A Surprising Pearl

The Pacific Rim Library (PRL) is an initiative of the Pacific Rim Digital Library Association (PRDLA). The project began in 2006 using the OAI-PMH paradigm and now holds over 300,000 records harvested from OAI data provider libraries around the Pacific. PRL's goal is to enable the sharing of digital collections amongst PRDLA members and the world, but greater unexpected benefits have been discovered. Through mirroring their metadata, PRL increases the chance that their data will be discovered in Google and other general search engines. With its many disparate collections, PRL is not a repository for traditional information discovery and retrieval. Initially users will bounce from a Google hit to the PRL metadata record in Hong Kong and then begin an intensive search on the original site which hosts the full digital object, in Vancouver, Honolulu, Wuhan, Singapore, or other PRDLA member location.     

Bayesian credible intervals for response surface optima

In response surface methodology, one is usually interested in estimating the optimal conditions based on a small number of experimental runs which are designed to optimally sample the experimental space. Typically, regression models are constructed from the experimental data and interrogated in order to provide a point estimate of the independent variable settings predicted to optimize the response. Unfortunately, these point estimates are rarely accompanied with uncertainty intervals. Though classical frequentist confidence intervals can be constructed for unconstrained quadratic models, higher order, constrained or nonlinear models are often encountered in practice. Existing techniques for constructing uncertainty estimates in such situations have not been implemented widely, due in part to the need to set adjustable parameters or because of limited or difficult applicability to constrained or nonlinear problems. To address these limitations a Bayesian method of determining credible intervals for response surface optima was developed. The approach shows good coverage probabilities on two test problems, is straightforward to implement and is readily applicable to the kind of constrained and/or nonlinear problems that frequently appear in practice.     

A counterexample to Beder's conjectures about Hadamard matrices

In this note we provide a counterexample which resolves conjectures about Hadamard matrices made in this journal. Beder [1998. Conjectures about Hadamard matrices. Journal of Statistical Planning and Inference 72, 7–14] conjectured that if H$\mathit{H}$ is a maximal m×n$m\times n$ row-Hadamard matrix then m is a multiple of 4; and that if n   is a power of 2 then every row-Hadamard matrix can be extended to a Hadamard matrix. Using binary integer programming we obtain a maximal 13×32$13\times 32$ row-Hadamard matrix, which disproves both conjectures. Additionally for n being a multiple of 4 up to 64, we tabulate values of m   for which we have found a maximal row-Hadamard matrix. Based on the tabulated results we conjecture that a m×n$m\times n$ row-Hadamard matrix with m?n-7$m?n-7$ can be extended to a Hadamard matrix.