Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: RE: -locpoly-

From   "D.Christodoulou" <>
Subject   Re: st: RE: -locpoly-
Date   Tue, 01 Jun 2004 12:33:41 +0100

Dear Statalisters,

Following Nick's reply on my previous posting I was not able to locate the
Fan, Fan, and Gijbels,  "Local Polynomial Modelling and Its Applications",
in Monographs of Statistics and Applied Probability, Chapman and Hall/CRC.

(1) I would like ask the authors of the -locpoly- command and whoever is
familiar with the command whether the following references are sufficient
in explaining the algorithm of -locoply-:

Hardle (1990), Applied Nonparametric Regression, Econometric Society
Monographs, Cambridge Press.

Cleveland (1979), "Robust Locally Weighted Regression and Smoothing
Scatterplots", Journal of the American Statistical Association, Vol.74,
No.368, pp.829-836.

Muller (1987), "Weighted Local Regression and Kernel Methods for
Nonparametric Fitting", Journal of the American Statistical Association,
Vol.82, No.397, pp.231-238.

(2) Has anyone written something on nonparametric smoothing for
higher-dimensions, i.e. more than one explanatory variables ?

(3) How can I statistically compare (given a confidence level) the
parametric fit "y-hat" with the non-parametric fit "conditional-y-mean" ?
Has anyone written a statistical test such as the one described in
Hardle(1990, ch.9.3) for the qsuared deviation between the two measures ?

many thanks in advance,

Nick Cox wrote:
> -locpoly- is a user-written program, except
> that the authors are all also StataCorp employees
> currently enjoying Memorial Day weekend: see
> -locpoly- was written up in the Stata Journal
> 4(3) 2003, and that write-up includes an array
> of references. One is the monograph by Fan and Gijbels
> on local polynomial modelling from Chapman and Hall
> (now CRC Press).
> -locpoly- is, I think, not accurately described as an
> extension of the user-written -kernreg-. It is
> a completely independent program. Both show
> some influence of -kdensity-: that's where the
> relationship lies if there is one.
> Confidence intervals do not appear to be built
> in. No doubt someone somewhere has written how
> to do this, with some extra work.
> Nick
> > Is it possible to construct and graph confidence intervals
> > with the -locpoly-
> > nonparametric regression smooth?
> >
> > Also, the -locpoly- help file does not point to any
> > references. Is the -locpoly-
> > an extension of the -kernreg- but with a polynomial fit instead?
> *
> *   For searches and help try:
> *
> *
> *

Dimitris Christodoulou
School for Business and Regional Development
University of Wales, Bangor
Hen Coleg
LL57 2DG Bangor
*   For searches and help try:

© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index