Resampling

About: Resampling is a research topic. Over the lifetime, 5428 publications have been published within this topic receiving 242291 citations.

Survey of resampling techniques for improving classification performance in unbalanced datasets.

Abstract: A number of classification problems need to deal with data imbalance between classes. Often it is desired to have a high recall on the minority class while maintaining a high precision on the majority class. In this paper, we review a number of resampling techniques proposed in literature to handle unbalanced datasets and study their effect on classification performance.

Bootstrapped Confidence Intervals as an Approach to Statistical Inference

Abstract: Confidence intervals are in many ways a more satisfactory basis for statistical inference than hypothesis tests. This article explains a simple method for using bootstrap resampling to derive confidence intervals. This method can be used for a wide variety of statistics—including the mean and median, the difference of two means or proportions, and correlation and regression coefficients. It can be implemented by an Excel spreadsheet, which is available to readers on the Web. The rationale behind the method is transparent, and it relies on almost no sophisticated statistical concepts.

An Introduction to Bootstrap Methods with Applications to R

Abstract: A comprehensive introduction to bootstrap methods in the R programming environmentBootstrap methods provide a powerful approach to statistical data analysis, as they have more general applications than standard parametric methods. An Introduction to Bootstrap Methods with Applications to R explores the practicality of this approach and successfully utilizes R to illustrate applications for the bootstrap and other resampling methods. This book provides a modern introduction to bootstrap methods for readers who do not have an extensive background in advanced mathematics. Emphasis throughout is on the use of bootstrap methods as an exploratory tool, including its value in variable selection and other modeling environments.The authors begin with a description of bootstrap methods and its relationship to other resampling methods, along with an overview of the wide variety of applications of the approach. Subsequent chapters offer coverage of improved confidence set estimation, estimation of error rates in discriminant analysis, and applications to a wide variety of hypothesis testing and estimation problems, including pharmaceutical, genomics, and economics. To inform readers on the limitations of the method, the book also exhibits counterexamples to the consistency of bootstrap methods.An introduction to R programming provides the needed preparation to work with the numerous exercises and applications presented throughout the book. A related website houses the book's R subroutines, and an extensive listing of references provides resources for further study.Discussing the topic at a remarkably practical and accessible level, An Introduction to Bootstrap Methods with Applications to R is an excellent book for introductory courses on bootstrap and resampling methods at the upper-undergraduate and graduate levels. It also serves as an insightful reference for practitioners working with data in engineering, medicine, and the social sciences who would like to acquire a basic understanding of bootstrap methods.

Local likelihood smoothing of sample extremes

Abstract: Trends in sample extremes are of interest in many contexts, an example being environmental statistics. Parametric models are often used to model trends in such data, but they may not be suitable for exploratory data analysis. This paper outlines a semiparametric approach to smoothing sample extremes, based on local polynomial fitting of the generalized extreme value distribution and related models. The uncertainty of fits is assessed by using resampling methods. The methods are applied to data on extreme temperatures and on record times for the women's 3000 m race.

Improvement Strategies for Monte Carlo Particle Filters

Abstract: The particle filtering field has seen an upsurge in interest over recent years, and accompanying this upsurge several enhancements to the basic techniques have been suggested in the literature. In this paper we collect a group of these developments that seem to be particularly important for time series applications and give a broad discussion of the methods, showing the relationships between them. We firstly present a general importance sampling framework for the filtering/smoothing problem and show how the standard techniques can be obtained from this general approach. In particular, we show that the auxiliary particle filtering methods of (Pitt and Shephard: this volume) fall into the same general class of algorithms as the standard bootstrap filter of (Gordon et al. 1993). We then develop the ideas further and describe the role of MCMC resampling as proposed by (Gilks and Berzuini: this volume) and (MacEachern, Clyde and Liu 1999). Finally, we present a generalisation of our own in which MCMC resampling ideas are used to traverse a sequence of ‘bridging’ densities which lie between the prediction density and the filtering density. In this way it is hoped to reduce the variability of the importance weights by attempting a series of smaller, more manageable moves at each time step.