Nonparametric Functional Estimation

Nonparametric Functional Estimation is a compendium of papers, written by experts, in the area of nonparametric functional estimation. This book attempts to be exhaustive in nature and is written both for specialists in the area as well as for students of statistics taking courses at the postgraduate level. The main emphasis throughout the book is on the discussion of several methods of estimation and on the study of their large sample properties. Chapters are devoted to topics on estimation of density and related functions, the application of density estimation to classification problems, and the different facets of estimation of distribution functions. Statisticians and students of statistics and engineering will find the text very useful.

About three years ago, an idea was discussed among some colleagues in the Division of Statistics at the University of California, Davis, as to the possibility of holding an international conference, focusing exclusively on nonparametric curve estimation. The fruition of this idea came about with the enthusiastic support of this project by Luc Devroye of McGill University, Canada, and Peter Robinson of the London School of Economics, UK. The response of colleagues, contacted to ascertain interest in participation in such a conference, was gratifying and made the effort involved worthwhile. Devroye and Robinson, together with this editor and George Metakides of the University of Patras, Greece and of the European Economic Communities, Brussels, formed the International Organizing Committee for a two week long Advanced Study Institute (ASI) sponsored by the Scientific Affairs Division of the North Atlantic Treaty Organization (NATO). The ASI was held on the Greek Island of Spetses between July 29 and August 10, 1990. Nonparametric functional estimation is a central topic in statistics, with applications in numerous substantive fields in mathematics, natural and social sciences, engineering and medicine. While there has been interest in nonparametric functional estimation for many years, this has grown of late, owing to increasing availability of large data sets and the ability to process them by means of improved computing facilities, along with the ability to display the results by means of sophisticated graphical procedures.

Topics emphasized include nonparametric density estimation as an exploratory device plus the deeper models to which the exploratory analysis points, multi-dimensional data analysis, and analysis of remote sensing data, cancer progression, chaos theory, epidemiological modeling, and parallel based algorithms. New methods discussed are quick nonparametric density estimation based techniques for resampling and simulation based estimation techniques not requiring closed form solutions.

Developed from lecture notes and ready to be used for a course on the graduate level, this concise text aims to introduce the fundamental concepts of nonparametric estimation theory while maintaining the exposition suitable for a first approach in the field.

This book gives a systematic, comprehensive, and unified account of modern nonparametric statistics of density estimation, nonparametric regression, filtering signals, and time series analysis. The companion software package, available over the Internet, brings all of the discussed topics into the realm of interactive research. Virtually every claim and development mentioned in the book is illustrated with graphs which are available for the reader to reproduce and modify, making the material fully transparent and allowing for complete interactivity.

Modern apparatuses allow us to collect samples of functional data, mainly curves but also images. On the other hand, nonparametric statistics produces useful tools for standard data exploration. This book links these two fields of modern statistics by explaining how functional data can be studied through parameter-free statistical ideas. At the same time it shows how functional data can be studied through parameter-free statistical ideas, and offers an original presentation of new nonparametric statistical methods for functional data analysis.

This book presents a unified approach on nonparametric estimators for models of independent observations, jump processes and continuous processes. New estimators are defined and their limiting behavior is studied. From a practical point of view, the book expounds on the construction of estimators for functionals of processes and densities, and provides asymptotic expansions and optimality properties from smooth estimators. It also presents new regular estimators for functionals of processes, compares histogram and kernel estimators, compares several new estimators for single-index models, and it examines the weak convergence of the estimators. Contents:IntroductionKernel Estimator of a DensityKernel Estimator of a Regression FunctionLimits for the Varying Bandwidths EstimatorsNonparametric Estimation of QuantilesNonparametric Estimation of Intensities for Stochastic ProcessesEstimation in Semi-Parametric Regression ModelsDiffusion ProcessesApplications to Time Series Readership: Advanced undergraduate and graduate students in mathematical statistics and computational statistics; researchers in mathematical or applied statistics; statisticians. Keywords:Kernel Estimation;Density;Regression;Intensity;Diffusion;Nonparametric;Weak Convergence;Ergodic Process;Intensity of Point Process;Kernel Estimation;Single-Index Models;Functional Time Series;Variable BandwidthKey Features:Covers a wide range of nonparametric models and presents new functional estimatorsContains a detailed presentation of the mathematical techniques for functional estimation with recent advances in their optimizationA valuable resource for scientific researchers who model observation dataReviews: “This book is useful for researchers interested in the study of asymptotic properties of different types of optimum estimators obtained through the method of kernels.” Mathematical Reviews