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: metrics
See also: sssm sir

Macro: sir1
Description: This macro is the taylored Sliced Inverse Regression method for the "sssm" macro. Here, the macro "sir1" builds a special slice which contains the cases such that y=0 (the value "0" for y symbolically indicates a missing value). Then, nbslices (by default, nbslices=5) other slices are made by splitting the range of the non missing y's into slices of nearly equal weight (i.e. such that they contain nearly the same number of cases). The classical algorithm of S.I.R. is then used with this special slicing.

Usage: {edr, eigen} = sir1 (matYX, nbslices)
Input:
matYX n x (p+1) matrix containing the value of the dependent variable in its first column, the values of the p explanatory variables in the last p columns, for a sample of size n.
nbslices optional scalar parameter, number of slices for the non missing y's cases. (By default, it is equal to 5)
Output:
edr p x p matrix which contains the estimated EDR directions (in columns), sorted according to the increasing values of their corresponding eigenvalues.
eigen p x 1 matrix which contains the corresponding eigenvalues sorted according to the increasing values.

Library: metrics
See also: sssm sir

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

Author: JS, 960916
(C) MD*TECH Method and Data Technologies, 17.8.2000