Naive Bayes classifier

E577500

machine learning model probabilistic classifier supervised learning algorithm

A Naive Bayes classifier is a simple probabilistic machine learning model that applies Bayes’ theorem under strong independence assumptions between features to perform fast and effective classification.

Try in SPARQL Jump to: Statements Referenced by

Statements (52)

Predicate	Object
instanceOf	machine learning model ⓘ probabilistic classifier ⓘ supervised learning algorithm ⓘ
advantage	low computational cost ⓘ scales well to large datasets ⓘ works with small training datasets ⓘ
assumes	conditional independence of features given the class ⓘ
basedOn	Bayes' theorem NERFINISHED ⓘ
commonImplementation	R packages ⓘ Weka NERFINISHED ⓘ scikit-learn NERFINISHED ⓘ
comparedTo	decision trees ⓘ logistic regression ⓘ support vector machines ⓘ
computes	posterior probability of each class ⓘ
field	machine learning ⓘ pattern recognition ⓘ statistics ⓘ
hasVariant	Bernoulli Naive Bayes NERFINISHED ⓘ Categorical Naive Bayes NERFINISHED ⓘ Complement Naive Bayes NERFINISHED ⓘ Gaussian Naive Bayes NERFINISHED ⓘ Multinomial Naive Bayes NERFINISHED ⓘ
isKnownFor	fast prediction ⓘ fast training ⓘ good performance on high-dimensional data ⓘ robustness to irrelevant features ⓘ simplicity ⓘ
isUsedFor	document categorization ⓘ medical diagnosis ⓘ recommendation systems ⓘ sentiment analysis ⓘ spam filtering ⓘ text classification ⓘ
limitation	performance can degrade with highly correlated features ⓘ probability estimates can be poorly calibrated ⓘ strong independence assumption may be violated ⓘ
oftenUses	Laplace smoothing NERFINISHED ⓘ additive smoothing ⓘ maximum likelihood estimation ⓘ
output	class label ⓘ class posterior probabilities ⓘ
predictionComplexity	linear in number of features and classes ⓘ
requires	estimation of class prior probabilities ⓘ estimation of conditional feature distributions ⓘ
trainingComplexity	linear in number of samples and features ⓘ
typicalDecisionRule	maximum a posteriori decision rule ⓘ
typicalFeatureModel	Bernoulli distribution for binary features ⓘ Gaussian distribution for continuous features ⓘ multinomial distribution for count features ⓘ
uses	likelihood of features given class ⓘ prior probabilities of classes ⓘ

Referenced by (1)

Full triples — surface form annotated when it differs from this entity's canonical label.

Bayes’ theorem → usedIn → Naive Bayes classifier ⓘ