Averaged One-Dependence Estimators

Synonyms

AODE

Definition

Averaged one-dependence estimators is a semi-naive Bayesian Learning method. It performs classification by aggregating the predictions of multiple one-dependence classifiers in which all attributes depend on the same single parent attribute as well as the class.

Classification with AODE

An effective approach to accommodating violations of naive Bayes’ attribute independence assumption is to allow an attribute to depend on other non-class attributes. To maintain efficiency it can be desirable to utilize one-dependence classifiers, such as Tree Augmented Naive Bayes (TAN), in which each attribute depends upon the class and at most one other attribute. However, most approaches to learning with one-dependence classifiers perform model selection, a process that usually imposes substantial computational overheads and substantially increases variance relative to naive Bayes.

AODE avoids model selection by averaging the predictions of multiple one-dependence...