[source]

Gaussian Naive Bayes#

Gaussian Naive Bayes is a version of the Naive Bayes classifier for continuous features. It places a probability density function over the input features on a class basis and uses Bayes' Theorem to derive the class probabilities. In addition to feature independence, Gaussian NB comes with the additional assumption that all features are normally (Gaussian) distributed.

Interfaces: Estimator, Learner, Online, Probabilistic, Persistable

Data Type Compatibility: Continuous

Parameters#

# Param Default Type Description
1 priors null array The class prior probabilities as an associative array with class labels as keys and the prior probabilities as values. If null, then the learner will compute these values from the training set.

Additional Methods#

Return the class prior probabilities:

public priors() : ?array

Return the running mean of each feature column for each class:

public means() : ?array

Return the running variance of each feature column for each class:

public variances() : ?array

Example#

use Rubix\ML\Classifiers\GaussianNB;

$estimator = new GaussianNB([
    'benign' => 0.9,
    'malignant' => 0.1,
]);

References#

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