Implementation of the Recursive Naive Bayes Learner for sequence classifier
###Keyword
Weka, Recursive Naive Bayes, Decision tree, Multinomial event model, sequence classifier
A java class for building and using a recursive Naive Bayes classifier for sequence classification. RNBL-MN is a tree of Naive Bayes classifiers, where each node is a NB classifier based on a multinominal event model.
The RNBL-MN is shown to outperforms C4.5 decision tree learner, and yields accuracies comparable to a SVM using similar information.
##Reference For more information see,
Dae-Ki Kang, Adrian Silvescu, Vasant Honavar "RNBL-MN: A Recursive Naive Bayes Learner for Sequences Classification" PAKDD'06.
This project relys on the Weka 3.6 NaiveBayesMultinominal classifiers and other assistant functions.
I add the C4.5 decision tree method in the evaluation to compare with RNBL-MN. The C4.5 takes significantly longer time to run, roughly 20 seconds for one fold. I conduct 10-fold cross validation on the 10 dataset, which is roughly 30 to 40 minues.