next up previous contents
Next: Ensemble learners Up: Practical details Previous: Accuracy measurement   Contents

Backend learners

TClass relies on a propositional learner at the backend. We consider four propositional learners: J48 (the Weka equivalent of c4.5), PART (the Weka equivalent of c4.5rules), IB1 (nearest neighbour algorithm) and Naive Bayes (with the Laplace adjustment to prevent zero probabilities and Fayyad-Irani [FI93] based discretisation).



Mohammed Waleed Kadous 2002-12-10