决策树、朴素贝叶斯和朴素贝叶斯树的比较

摘要

In this paper, we used the normal dataset 'adult' to compare the runing 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.%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.