Scaling up the accuracy of averaged one-dependence estimators with decision tree-based attribute weighted

Averaged One-Dependence Estimators (AODE) is a most effective improved naive Bayes (NB) algorithm based on probabilistic classification learning technique. It addresses the attribute independence assumption of naive Bayes by averaging all of the dependence estimators. Researchers have proposed out many effective methods to improve the performance of AODE, such as attribute weighted method, backwards sequential elimination method, lazy elimination method and so on. In this paper, our research is focused on weighted method. We firstly present a simple filter method for setting attribute weights of AODE and then present an improved algorithm called Decision Tree-Based Attribute Weighted Averaged One-Dependence Estimator, simply DTWAODE. In DTWAODE, the weight for an attribute is set according to its depth in a decision tree which is built on the training samples. We experimentally tested DTWAODE in Weka system, using the whole 36 standard UCI data sets and the experimental results show that our new algorithm performs better than AODE.