| Library: | smoother |
| See also: | regci regcb regxest regxbwsel lpregest regestp |
| Macro: | regest | |
| Description: | computes the Nadaraya-Watson estimator for univariate regression. The computation uses WARPing. |
| Usage: | mh = regest(x {,h {,K} {,d} }) | |
| Input: | ||
| x | n x 2, the data. In the first column the independent, in the second column the dependent variable. | |
| h | scalar, bandwidth. If not given, 20% of the range of x[,1] is used. | |
| K | string, kernel function on [-1,1] or Gaussian kernel "gau". If not given, the Quartic kernel "qua" is used. | |
| d | scalar, discretization binwidth. d must be smaller than h. If not given, the minimum of h/3 and (max(x[,1])-min(x[,1]))/100 is used. | |
| Output: | ||
| mh | m x 2 matrix, the first column is a grid and the second column contains the regression estimate on that grid. | |
library("smoother")
library("plot")
;
x = 4.*pi.*(uniform(200)-0.5) ; independent variable
m = cos(x) ; true function
e = uniform(200)-0.5 ; error term
x = x~(m+e)
;
mh = regest(x,1) ; estimate function
;
mh = setmask(mh, "line","blue")
m = setmask(sort(x[,1]~m) , "line","black","thin")
plot(x,mh,m)
The Nadaraya-Watson regession estimate (blue) using Quartic kernel and bandwidth h=1 and the true regression function (thin black) are pictured.
library("smoother")
library("plot")
;
x = read("motcyc") ; read motorcycle data
mhe = regest(x,3,"epa") ; estimate function
mhu = regest(x,2,"uni") ; estimate function
;
mhe= setmask(mhe,"line","green")
mhu= setmask(mhu,"line","red")
plot(mhe,mhu) ; graph functions
The Nadaraya-Watson regession estimates using Epanechnikov kernel (green) and Uniform kernel (red) are pictured.
| Library: | smoother |
| See also: | regci regcb regxest regxbwsel lpregest regestp |