Prospect of Nonparametric Modeling
Lists
Only you see this
Abstract
Prospects of nonparametric modeling * Jianqing Fan Department of Statistics, U n i v e r s i t y of C a l i f o r n i a , Los Angeles, C A 90095 Abstract This article briefly summarizes key developments in nonparametric function estimation over the past three decades. It highlights some potentially fruitful areas of future research on non- parametric modeling and its applications. Introduction M o d e r n c o m p u t i n g facilities allow statisticians to explore fine d a t a structures that were u n i m a g - inable two decades ago. D r i v e n by m a n y sophisticated applications, demanded by the need of nonlinear m o d e l i n g a n d fueled by m o d e r n c o m p u t i n g power, m a n y c o m p u t a t i o n a l l y intensive d a t a - a n a l y t i c m o d e l i n g techniques have been invented to exploit possible h i d d e n structures a n d to reduce m o d e l i n g biases of t r a d i t i o n a l parametric methods. These d a t a - a n a l y t i c approaches are also referred to as nonparametric techniques. For a n i n t r o d u c t i o n to these nonparametric techniques, see the books by Devroye a n d G y o r f i (1985), S i l v e r m a n (1986), E u b a n k (1988), M i i l l e r (1988), G y o r f i , Haxdle, S a r d a a n d V i e w (1989), Hastie a n d T i b s h i r a n i (1990), W a h b a (1990), Scott (1992), G r e e n a n d S i l v e r m a n (1994), W a n d a n d Jones (1995), F a n a n d G i j b e l s (1996), Simonoff (1996), B o w - m a n a n d A z z a l i n i (1997), H a r t (1997), R a m s a y a n d S i l v e r m a n (1997), O g d e n (1997), B o s q (1998), E f r o m o v i c h (1999), V i d a k o v i c (1999), among others. A n a i m of nonparametric techniques is to reduce possible m o d e l i n g biases of parametric models. S u c h parametric models are simple a n d convenient linear models to facilitate c o m p u t a t i o n a l expe- diency before 1980s'. T h e y are t y p i c a l l y not derived from p h y s i c a l laws a n d c a n not be expected to fit a l l d a t a well. A n erroneous parametric model c a n create excessive m o d e l i n g biases a n d leads to w r o n g conclusions. N o n p a r a m e t r i c techniques intend to fit a m u c h larger class of models to reduce m o d e l i n g biases. T h e y allow d a t a to search appropriate nonlinear forms that best describe the available d a t a . T h e y also provide useful tools for parametric nonlinear m o d e l i n g a n d for m o d e l diagnostics. Over the past three decades, intensive efforts have been devoted to nonparametric function estimation. M a n y new nonparametric models have been introduced a n d a vast array of new tech- niques have been invented. M a n y new phenomena have been unveiled a n d deep insights have been gained. T h e field of nonparametric m o d e l i n g has progressed steadily a n d d y n a m i c a l l y . T h i s t r e n d w i l l continue for decades to come. W i t h the advance of i n f o r m a t i o n a n d technology, more a n d more complicated d a t a m i n i n g problems emerge. T h e research a n d applications of d a t a - a n a l y t i c techniques w i l l be proven to be even more f r u i t f u l i n the next m i l l e n n i u m . T h e field of nonparametric m o d e l i n g is vast. It has taken many books to describe a part of the art. Indeed, most of parametric models have their nonparametric counterpart. It is impossible to * Fan's research was partially supported by N S F grant DMS-9804414 and a grant from University of California at Los Angeles.
BibTeX
@inproceedings{fan2011prospect,
author = {Jianqing Fan},
title = {{Prospect of Nonparametric Modeling}},
booktitle = {eScholarship},
year = {2011},
}Loading related content...