Bayesian and probabilistic learning algorithms.
This module provides classifiers and regressors based on Bayes' theorem, graphical models, and kernel-based probabilistic methods. It includes standard implementations of Naive Bayes, Bayesian Networks, and Gaussian Processes.
Algorithms
-
NaiveBayesClassifier: Gaussian and kernel-indexed probabilistic classifier.
-
NaiveBayesMultinomialClassifier: Specialized for discrete/text classification.
-
CategoricalNBClassifier: Naive Bayes for nominal / integer-coded categorical features.
-
GaussianProcessesRegressor: Bayesian non-parametric regression with
uncertainty estimation.
Estimators
The module also provides pluggable probability estimators used by the above models for modeling continuous and discrete feature distributions.