决策树、朴素贝叶斯和朴素贝叶斯树的比较

Comparison of Decision Tree, Native Bayesion and Natvie Bayesion Tree

利用数据挖掘工具Weka,在常态数据集adult上进行实验,从时间、正确率、误差率三个指标对比分析J48(C4.5决策树),朴素贝叶斯分类器,朴素贝叶斯树(NBTree)三种分类算法.结论为:在内存充足,时间要求不高的情况下,使用朴素贝叶斯树(NBTree)能获得更高的正确率和错误率;J48则是一种折中的方案;朴素贝叶斯分类器完成时间最短,但正确率和错误率为三种算法中最差.

In this paper, we used the normal dataset ＇adult＇ to compare the tuning time, correct rate and error rate of c4.5 decition tree, native bayesion and native bayesion tree, with the data mining tool called weka. The result shows that： we can choose the NB Tree when the memory is big enough and the running time is undemanding; and choose natvie bayesion when the job should be done quickly; and the c4.5（j48） decision tree is an compromising approach.