Описание
Chapters 4 , 5 , 8 , and 9 of this text. For time series and other dependent data, the
moving block bootstrap has become the method of choice and other block bootstrap methods have been developed. Other bootstrap techniques for dependent
data include transformation - based bootstrap (primarily the frequency domain
bootstrap) and the sieve bootstrap. Lahiri has been one of the pioneers at developing bootstrap methods for dependent data, and his text Lahiri ( 2003a ) covers
these methods and their statistical properties in great detail along with some
results for the IID case. To my knowledge, it is the only major bootstrap text
with extensive theory and applications from 2001 to 2003.
Since the fi rst edition of my text, I have given a number of short courses
on the bootstrap using materials from this and other texts as have others. In
the process, new examples and illustrations have been found that are useful
in a course text. The bootstrap is also being taught in many graduate school
statistics classes as well as in some elementary undergraduate classes. The
value of bootstrap methods is now well established.
The intention of the fi rst edition was to provide a historical perspective to
the development of the bootstrap, to provide practitioners with enough applications and references to know when and how the bootstrap can be used and
to also understand its pitfalls. It had a second purpose to introduce others to
the bootstrap, who may not be familiar with it, so that they can learn the basics
and pursue further advances, if they are so interested. It was not intended to
be used exclusively as a graduate text on the bootstrap. However, it could be
used as such with supplemental materials, whereas the text by Davison and
Hinkley ( 1997 ) is a self - contained graduate - level text. In a graduate course,
this book could also be used as supplemental material to one of the other fi ne
texts on bootstrap, particularly Davison and Hinkley ( 1997 ) and Efron and
Tibshirani ( 1993 ). Student exercises were not included; and although the
number of illustrative examples is increased in this edition, I do not include
exercises at the end of the chapters.
For the most part the fi rst edition was successful, but there were a few
critics. The main complaints were with regard to lack of detail in the middle
and latter chapters. There, I was sketchy in the exposition and relied on other
reference articles and texts for the details. In some cases the material had too
much of an encyclopedic fl avor. Consequently, I have expanded on the description of the bootstrap approach to censored data in Section 8.4 , and to p - value
adjustment in Section 8.5 . In addition to the discussion of kriging in Section
8.1 , I have added some coverage of other results for spatial data that is also
covered in Lahiri ( 2003a ).
There are no new chapters in this edition and I tried not to add too many
pages to the original bibliography, while adding substantially to Chapters 4
(on regression), 5 (on forecasting and time series), 8 (special topics), and 9
(when the bootstrap fails and remedies) and somewhat to Chapter 3 (on
hypothesis testing and confi dence intervals). Applications in the pharmaceutical industry such as the use of bootstrap for estimating individual and population bioequivalence are also included in a new Section 8.6 .
x preface to second edition
Детали
- Год издания
- 2007
- Format