Monday, July 5, 2010

the performance of naive Bayesian network

1, does more attribute definitely improve the performance of naive Bayesian classifier?
continue the example by assuming the first LR is derived from age, and the second from weight and the predicted value is the chance of a disease.

2, reference for feature selection
Learning Augmented Bayesian Classifiers: A Comparison of Distribution-based and Classification-based Approaches.
Feature subset selection as search with probabilistic estimates.
Scaling up the Naive Bayesian Classifier: Using Decision Trees for Feature Selection.
Induction of Selective Bayesian Classifiers.
Bayesian Network Classfiers


3, an interesting thesis on ml
Attribute Interactions in Machine Learning

4, ML book
Introduction to Machine Learning

No comments: