Mathematica 9 is now available
Services & Resources / Wolfram Forums / MathGroup Archive
-----

MathGroup Archive 2010

[Date Index] [Thread Index] [Author Index]

Search the Archive

Non-parametric regression and density estimation package posted to

  • To: mathgroup at smc.vnet.net
  • Subject: [mg113286] Non-parametric regression and density estimation package posted to
  • From: BernieTheJet <berniethejet at gmail.com>
  • Date: Fri, 22 Oct 2010 01:35:16 -0400 (EDT)

A number of years ago I wrote a package for non-parametric regression
and density estimation.  I have finally got around to posting it on
the Wolfram Library site:

http://library.wolfram.com/

Regards,

Bernard Gress

Here is the old blurb:


NonParametrix.m is a Mathematica package that provides many of the
basic (as well as a few advanced) functions often used in
nonparametric econometrics and statistics, as described in, for
example, Pagan and Ullah (1999), Silverman (1986), or H=E4rdle (1989).

NPKDense

Provides nonparametric density estimation of multivariate data with
user-defined kernels. Includes very fast routines for standard kernels
such as the normal, Epanechinikov, and the uniform.

NPRegress

Uses optimized, high-speed routines to fit an arbitrary order,
nonparametric local polynomial regression estimation on multivariate
data of any dimension, with pre-defined or user-defined kernels. Also
estimates slopes, regression residuals, confidence intervals, and can
output an Interpolating function of the Fits to use elsewhere. Also
includes high-speed local polynomial, generalized least squares (GLS)
routines.

CrossValidatedH

Finds the optimal window width ('h') for multivariate nonparametric
regressions. Includes mean square error, root mean square error,
median absolute error, mean absolute percentage error, and median
prediction error loss functions. Allows the user to select many of the
parameters of the cross-validation routine. Also allows for the use of
arbitrary kernels, albeit at much slower speeds.

Other Assorted Functions

SilvermanH calculates the well-known asymptotic kernel window width
for data of arbitrary dimensions. FastNPDensity provides very quick
and dirty univariate density estimation when the user is mostly
interested in quick visualization of data. MultipleKDensityPlot
provides quick and dirty estimation of multiple densities, color coded
for easy data visualization.

NonParametrix.m _WAS_ available at http://student.ucr.edu/~gressb01/index.html

<\End Blurb\>


  • Prev by Date: Speeding up InterpolatingFunction
  • Next by Date: Error with VoronoiDiagram[] function
  • Previous by thread: Re: Speeding up InterpolatingFunction
  • Next by thread: Error with VoronoiDiagram[] function