基于决策树的网络流量分类方法

摘要

针对传统流量分类方法(基于端口和有效载荷)分类不可靠的问题,提出基于C4.5决策树算法,根据训练集中属性的信息增益比率构建分类模型,按属性对测试数据集进行预测,通过查找分类模型实现对网络流量的分类.在公开数据集和自己采集的数据集上进行实验,结果表明,采用C4.5决策树算法对网络流量分类,平均分类精度为93％,单类别分类精度均在90％以上,能有效地实现对网络流量应用类型的识别.%Aiming at the problem of instability in traditional traffic classification methods,a traffic classification method based on G4. 5 decision tree is proposed,winch establishes models on the information gain ratio from the training set. Classifier is tested by attributes on test dataset.as well as network traffic is classified by searching classification models. Experiments show that the overall accuracy of our method achieves more than 93% , and the accuracy of single class is more than 90% on open dataset. So ihe method is effective for classifying various kinds of traffic.