决策树下引入残差数据合并的冗余数据挖掘  

Redundant Data Mining Based on Residual Data Merging in Decision Tree

作　　者：王倩[1] 

机构地区：[1]天津职业大学网络中心,天津300402

出　　处：《科技通报》2014年第6期155-157,共3页Bulletin of Science and Technology

基　　金：天津市教育科学"十二五"规划课题(VE4035)

摘　　要：提出采用残差数据合并技术的冗余数据优化挖掘算法,利用训练集建立决策树模型,引入C4.5决策树模型进行冗余数据主特征建模,在主分量特征决策树下,引入残差数据合并技术,设定数据残差特征伴随追踪模式,把传统方法中用于滤除的数据信息进行拼接伴随追踪定位,实现了冗余数据特征的优化挖掘。把方法应用到网络流量时间序列数据处理中实现网络异常监测,仿真实验表明,新的数据挖掘算法能有效提取到冗余数据特征作为有用检测特征,数据挖掘效率大幅提高,有效促进了海量数据隐藏特征的挖掘和应用,设计的网络流量监测软件能提高网络管理和监测实效性。An improved optimization data mining algorithm based on redundant data merging technology was proposed. The training set was used to build the decision tree model, the C4.5 decision tree model was used for redundant data main fea-ture modeling. The accompanied tracking model of residual feature was set, and the information was used for tracking and positioning with data splicing. The optimization of redundant data mining was realized finally. It was applied into the net-work traffic anomaly detection, simulation result shows that improved method can extract the effective redundant data fea-ture as useful feature, and data mining efficiency is improved greatly. It can promote the massive data mining development with using the hidden features. And the designed network traffic monitoring software can improve the effectiveness of net-work management and monitoring.

关 键 词：决策树 残差数据 数据挖掘 网络流量 

分 类 号：TP392[自动化与计算机技术—计算机应用技术]