Next: dasteplook() Up: Multivariate Macros Help File Previous: daentervar()   Contents


daremovevar(var1 [,var2 ...] [,silent:T]), var1, var2 ... names or
  numbers of variables to be removed

Keywords: classification, discrimination, stepwise
daremovevar(j), where j is a positive integer, removes dependent
variable j as part of a stepwise dependent variable selection, usually
as one stage in stepwise discriminant analysis.  This is what is
sometimes called a "backward" step.  It is an error if variable j is
not an "in" variable (is already an "out" variable).

daremovevar() updates variable _DASTEPSTATE which encapsulates the
current state of the variable selection process and prints a report with
the new values of F-to-remove or F-to-remove, and their P-values.  See
topic '_DASTEPSTATE' and dastepsetup() for more details.  It returns a
copy of the updated _DASTEPSTATE as an invisible variable that can be
assigned but is not normally printed.

You normally choose which variable to remove as the variable with the
smallest value of F-to-remove as printed by macro dastepsetup(),
dastepstatus() or a preceding use of daremovevar().

daremovevar(varname), where varname is the quoted or unquoted name of a
variable in the model, does the same.  Thus if the original data matrix
had column labels "SepLen", "SepWid", "PetLen" and "PetWid", either
daremovevar(SepWid) or daremovevar("SepWid") would be equivalent to

daremovevar(j1, j2, ...) and daremovevar(varname1, varname2, ...)
successively remove several variables, printing a report after each is
removed.  The value returned is _DASTEPSTATE after all the variables
have been removed

daremovevar(j1 [,j2, ...], silent:T) and daremovevar(varname1
[,varname2,...], silent:T) do the same, except no report is printed.
This would normally be followed by dastepstatus() to print a report
after all the variables have been removed.

See also dastepstatus(), daentervar()  and dasteplook().

Gary Oehlert 2003-01-15