Name
bkwols — BKW scaled multicollinearity diagnostic
CALLING SEQUENCE
[condindex]=bkwols(arg1,
,argn)
PARAMETERS
- Input
x = independent variables matrix (nobs x nvar)
- Output
condindex = the condition number
DESCRIPTION
Computes BKW COND index of a x matrix (with variables studentized to achieve invariance with respect to their scale). Contrary to bkw_scale, this function does not provide the variance-decomposition matrix.
EXAMPLE
x = grand(200,5,'nor',0,1);
x(:,1) = ones(200,1);
x(:,3) = x(:,2) + grand(200,1,'nor',0,1)*0.05;
// create multioncolinearity between columns 2 and 3
condindex=bkwols(x)
// Useful mainly for programming purpose (since bkw does much more).
AUTHOR
Eric Dubois 2002/Emmanuel Michaux 2005