XploRe Help : linregres

Keywords - Function groups - @ A B C D E F G H I J K L M N O P Q R S T U V W X Y Z

Library:	stats
See also:	gls linreg linregfs

Quantlet:		linregres
Description:		linregres computes some residual analysis for a linear regression.

Reference(s):

Chatterjee, S. and Price, B. (1991), Regression Analysis by Example, Whiley, p. 83-91

Usage:	`{res,out} = linregres (x, y, yh)`
Input:
	`x`	n x p regressors
	`y`	n x 1
	`yh`	n x 1
Output:
	`xfs`	n x 4
	`out`	n x 2

Note:

The return values are res, which contains the residuals in the first column, the leverage in the second column, the standardized residuals in the third column and the Cook distance in the last column. out consists of vectors of zeros and ones indicating some influential points. The first column indicates all observations with a leverage larger than 2*p/n and the second column all points with a Cook distance larger than 1.

Example:

; loads the library stats library("stats") ; reset random generator randomize(0) ; generate x x = normal(100, 3) ; generate y y = 10*x[,3]+x[,1].*x[,2] ; do the forward selection {xfs,bfs}=linregfs(x, y, 0.05) ; compute residual number {res,out}=linregres(xfs, y, xfs*bfs) ; create a display fro plotting disp = createdisplay(2,2) ; show residual plots ; residuals leverage ; standardized residuals Cook distance show(disp, 1, 1, y~res[,1]) show(disp, 2, 1, y~res[,3]) show(disp, 1, 2, y~res[,2]) show(disp, 2, 2, y~res[,4])

Result:

shows the residual plots. From the standardized residuals
we find three points in the bottom with an absolute value
larger than 3. The leverage plot shows on the right and
the left sets of points with a leverage larger than 0.04
(thus they are influential). The Cook distance plot shows
no datapoints larger than 1 and we can not find outliers
in x.

Author: S. Klinke, 19970522