内容简介
Bootstrapmethodsarecomputer-intensivemethodsofstatisticalanalysis,whichusesimulationtocalculatestandarderrors,confidenceintervals,andsignificancetests.Themethodsapplyforanylevelofmodelling,andsocanbeusedforfullyparametric,semiparametric,andcompletelynonparametricanalysis.This1997bookgivesabroadandup-to-datecoverageofbootstrapmethods,withnumerousappliedexamples,developedinacoherentwaywiththenecessarytheoreticalbasis.Applicationsincludestratifieddata;finitepopulations;censoredandmissingdata;linear,nonlinear,andsmoothregressionmodels;classification;timeseriesandspatialproblems.Specialfeaturesofthebookinclude:extensivediscussionofsignificancetestsandconfidenceintervals;materialonvariousdiagnosticmethods;andmethodsforefficientcomputation,includingimprovedMonteCarlosimulation.Eachchapterincludesbothpracticalandtheoreticalexercises.S-Plusprogramsforimplementingthemethodsdescribedinthetextareavailablefromthesupportingwebsite.