Directed tree algorithm based on TAN classifiers
Jianlin Wang · Jisuanji gongcheng yu sheji · 2008
TAN(tree augmented Nave Bayes) takes the Nave Bayes classifier and adds edges to it,it is efficient extend of Nave Bayes.Since basic TAN learning algorithm choice tree structure rooted randomly,that makes it unable to express the dependence among attributes accurately.The directed dependence among attributes is set,and directed tree algorithm is added into TAN learning algorithm,then a new TAN algorithm is proposed — DTAN(directed tree augmented Nave Bayes).Finally,experiments is carried out to evaluate the DTAN and it is compared with basic TAN and Nave Bayes.The experimental results show that the accuracy of the DTAN is much higher than that of basic TAN and Nave Bayes on the big instance of datasets.