Density estimation has evolved enormously since the days of bar plots and histograms, but researchers and users are still struggling with the problem of the selection of the bin widths. This text explores a new paradigm for the data-based or automatic selection of the free parameters of density estimates in general so that the expected error is within a given constant multiple of the best possible error. The paradigm can be used in nearly all density estimates and for most model selection problems, both parametric and nonparametric. It is the first book on this topic. The text is intended for first-year graduate students in statistics and learning theory, and offers a host of opportunities for further research and thesis topics. Each chapter corresponds roughly to one lecture, and is supplemented with many classroom exercises. A one year course in probability theory at the level of Feller's Volume 1 should be more than adequate preparation. Gabor Lugosi is Professor at Universitat Pompeu Fabra in Barcelona, and Luc Debroye is Professor at McGill University in Montreal. In 1996, the authors, together with Lászlo Györfi, published the successful text, A Probabilistic Theory of Pattern Recognition with Springer-Verlag. Both authors have made many contributions in the area of nonparametric estimation.

Density estimation has evolved enormously since the days of bar plots and histograms, but researchers and users are still struggling with the problem of the selection of the bin widths. This book is the first to explore a new paradigm for the data-based or automatic selection of the free parameters of density estimates in general so that the expected error is within a given constant multiple of the best possible error. The paradigm can be used in nearly all density estimates and for most model selection problems, both parametric and nonparametric.

Density estimation has evolved enormously since the days of bar plots and histograms, but researchers and users are still struggling with the problem of the selection of the bin widths. This book is the first to explore a new paradigm for the data-based or automatic selection of the free parameters of density estimates in general so that the expected error is within a given constant multiple of the best possible error. The paradigm can be used in nearly all density estimates and for most model selection problems, both parametric and nonparametric.

This thesis explores combinatorial methods in random vector balancing, nonparametric estimation, and network inference. First, motivated by problems from controlled experiments, we study random vector balancing from the perspective of discrepancy theory, a classical topic in combinatorics, and give sharp statistical results along with improved algorithmic guarantees. Next, we focus on the problem of density estimation and investigate the fundamental statistical limits of coresets, a popular framework for obtaining algorithmic speedups by replacing a large dataset with a representative subset. In the following chapter, motivated by the problem of fast evaluation of kernel density estimators, we demonstrate how a multivariate interpolation scheme from finite-element theory based on the combinatorial-geometric properties of a certain mesh can be used to significantly improve the storage and query time of a nonparametric estimator while also preserving its accuracy. Our final chapter focuses on pedigree reconstruction, a combinatorial inference task of recovering the latent network of familial relationships of a population from its extant genetic data.

Now available in paperback, this book covers some recent developments in statistical inference. It provides methods applicable in problems involving nuisance parameters such as those encountered in comparing two exponential distributions or in ANOVA without the assumption of equal error variances. The generalized procedures are shown to be more powerful in detecting significant experimental results and in avoiding misleading conclusions.

The book provides a comprehensive treatment of statistical inference using permutation techniques. It features a variety of useful and powerful data analytic tools that rely on very few distributional assumptions. Although many of these procedures have appeared in journal articles, they are not readily available to practitioners.

By giving a detailed account of bootstrap methods and their properties for dependent data, this book provides illustrative numerical examples throughout. The book fills a gap in the literature covering research on re-sampling methods for dependent data that has witnessed vigorous growth over the last two decades but remains scattered in various statistics and econometrics journals. It can be used as a graduate level text and also as a research monograph for statisticians and econometricians.

This book presents a unique approach to stream data mining. Unlike the vast majority of previous approaches, which are largely based on heuristics, it highlights methods and algorithms that are mathematically justified. First, it describes how to adapt static decision trees to accommodate data streams; in this regard, new splitting criteria are developed to guarantee that they are asymptotically equivalent to the classical batch tree. Moreover, new decision trees are designed, leading to the original concept of hybrid trees. In turn, nonparametric techniques based on Parzen kernels and orthogonal series are employed to address concept drift in the problem of non-stationary regressions and classification in a time-varying environment. Lastly, an extremely challenging problem that involves designing ensembles and automatically choosing their sizes is described and solved. Given its scope, the book is intended for a professional audience of researchers and practitioners who deal with stream data, e.g. in telecommunication, banking, and sensor networks.

A fundamental statistical framework for the analysis of complex longitudinal data is provided in this book. It provides the first comprehensive description of optimal estimation techniques based on time-dependent data structures. The techniques go beyond standard statistical approaches and can be used to teach masters and Ph.D. students. The text is ideally suitable for researchers in statistics with a strong interest in the analysis of complex longitudinal data.