Predicate/Reducer and Probabilistic Predicates

The OpenCDS Predicate/Reducer is the core component of the DS2 Predicate/Reducer Architecture. Through the DS2 Predicate APIs it can utilize probabilistic predicates in addition to deterministic predicates built using OpenCDS rules.  Combined, the Predicate/Reducer and the probabilistic predicate implementations span three projects. 

Click on the project name to access the project's Bitbucket Repository, which contains documentation and build instructions.

 

Predicate Reducer

OpenCDS Predicate-Reducer called by CDA Web Service. Utilizes Deterministic Predicates and the NaiveBayesPredicate Probabilistic Predicate.

NaiveBayesPredicate

A SimpleProbabilisticPredicate implementation of the Naive Bayes classifier in Java. Can be used by the OpenCDS Predicate-Reducer, the Visualizer, or by any Java application.

RandomForestPredicate

A SimpleProbabilisticPredicate implementation using WEKA's Random Forest classifier. Can be used by the OpenCDS Predicate-Reducer, the Visualizer, or by any Java application. Despite its name, this predicate can actually utilize any WEKA classifier with binary attribute capability; and has also been tested with Naive Bayes, AODE, Bayes Network, and K-Nearest-Neighbors. Swapping out a classifier is as easy as training a new classifier in WEKA and saving the model.