Mercurial > hg > octave-max
changeset 14234:938a8d792c37
zscore.m: Return mean and std dev for Matlab compatibility. Update docstring.
author | Jordi Gutiérrez Hermoso <jordigh@octave.org> |
---|---|
date | Thu, 19 Jan 2012 12:46:33 -0500 |
parents | 358b59706656 |
children | 921b15c13adc |
files | scripts/statistics/base/zscore.m |
diffstat | 1 files changed, 36 insertions(+), 14 deletions(-) [+] |
line wrap: on
line diff
--- a/scripts/statistics/base/zscore.m +++ b/scripts/statistics/base/zscore.m @@ -17,23 +17,31 @@ ## <http://www.gnu.org/licenses/>. ## -*- texinfo -*- -## @deftypefn {Function File} {} zscore (@var{x}) -## @deftypefnx {Function File} {} zscore (@var{x}, @var{dim}) +## @deftypefn {Function File} {[@var{z}, @var{mu}, @var{sigma}] =} zscore (@var{x}) +## @deftypefnx {Function File} {[@var{z}, @var{mu}, @var{sigma}] =} zscore (@var{x}, @var{opt}) +## @deftypefnx {Function File} {[@var{z}, @var{mu}, @var{sigma}] =} zscore (@var{x}, @var{opt}, @var{dim}) ## If @var{x} is a vector, subtract its mean and divide by its standard -## deviation. +## deviation. If the standard deviation is zero, divide by 1 instead. +## The optional parameter @var{opt} determines the normalization to use +## when computing the standard deviation and is the same as the +## corresponding parameter for @code{std}. ## ## If @var{x} is a matrix, do the above along the first non-singleton -## dimension. -## If the optional argument @var{dim} is given, operate along this dimension. -## @seealso{center} +## dimension. If the third optional argument @var{dim} is given, operate +## along this dimension. +## +## The mean and standard deviation along @var{dim} are given in @var{mu} +## and @var{sigma} respectively. +## +## @seealso{mean, std, center} ## @end deftypefn ## Author: KH <Kurt.Hornik@wu-wien.ac.at> ## Description: Subtract mean and divide by standard deviation -function z = zscore (x, dim) +function [z, mu, sigma] = zscore (x, opt, dim) - if (nargin != 1 && nargin != 2) + if (nargin < 1 || nargin > 3 ) print_usage (); endif @@ -41,9 +49,17 @@ error ("zscore: X must be a numeric vector or matrix"); endif + if (nargin < 2) + opt = 0; + else + if (opt != 0 && opt != 1 || ! isscalar(opt)) + error("zscore: OPT must be empty, 0, or 1"); + endif + endif + nd = ndims (x); sz = size (x); - if (nargin != 2) + if (nargin < 3) ## Find the first non-singleton dimension. (dim = find (sz > 1, 1)) || (dim = 1); else @@ -57,11 +73,17 @@ if (n == 0) z = x; else - x = center (x, dim); # center also promotes integer to double for next line - z = zeros (sz, class (x)); - s = std (x, [], dim); + + if (isinteger (x)) + x = double (x); + endif + + mu = mean (x, dim); + sigma = std (x, opt, dim); + s = sigma; s(s==0) = 1; - z = bsxfun (@rdivide, x, s); + ## FIXME: Use normal broadcasting once we can disable that warning + z = bsxfun (@rdivide, bsxfun (@minus, x, mu), s); endif endfunction @@ -79,6 +101,6 @@ %!error zscore (['A'; 'B']) %!error zscore (1, ones(2,2)) %!error zscore (1, 1.5) -%!error zscore (1, 0) +%!error zscore (1, 1, 0) %!error zscore (1, 3)