Z Scale Standardizer#

A method of centering and scaling a dataset such that it has 0 mean and unit variance, also known as a Z-Score. Although Z-Scores are technically unbounded, in practice they mostly fall between -3 and 3 - that is, they are no more than 3 standard deviations away from the mean.

Interfaces: Transformer, Stateful, Elastic

Data Type Compatibility: Continuous


# Param Default Type Description
1 center true bool Should we center the sample dataset?

Additional Methods#

Return the means calculated by fitting the training set:

public means() : array

Return the variances calculated during fitting:

public variances() : array

Return the standard deviations calculated during fitting:

public stddevs() : array


use Rubix\ML\Transformers\ZScaleStandardizer;

$transformer = new ZScaleStandardizer(true);


  • T. F. Chan et al. (1979). Updating Formulae and a Pairwise Algorithm for Computing Sample Variances.