基于随机森林的无线通信数据多尺度离群点挖掘研究  

Study on Multi-scale Outlier Point Mining of Wireless Communication Data Based on Random Forest

作　　者：宋可可 郑含笑 宋浩浩 王思明 徐萌瑶 SONG Keke;ZHENG Hanxiao;SONG Haohao;WANG Siming;XU Mengyao(Shangqiu Institute of Technology,School of Information and Electronic Engineering,ShangQiu 476000,China)

机构地区：[1]商丘工学院信息与电子工程学院,河南商丘476000

出　　处：《通信电源技术》2025年第4期174-176,共3页Telecom Power Technology

摘　　要：为提高无线通信数据离群点挖掘的精度与效率,提出基于随机森林的无线通信数据多尺度离群点挖掘研究。首先,对无线通信数据进行降维处理,以减少数据冗余并保留关键信息。其次,基于降维后的数据,提取离群点特征,以突出异常数据的特性。最后,构建随机森林模型,计算预测残差,将其作为离群度的衡量指标,进而实现多尺度挖掘,并标记离群点。实验结果显示,应用该方法后成功将无线通信节点划分为不同的簇,进而精确地识别出离群点,且随着数据量的增加,挖掘执行时间始终保持在较低水平,证明了该方法具有较高的离群点挖掘效率。In order to improve the accuracy and efficiency of outlier mining in wireless communication data,a research on multi-scale outlier mining in wireless communication data based on random forest is proposed.Firstly,the dimension of wireless communication data is reduced to reduce data redundancy and retain key information.Secondly,based on the data after dimensionality reduction,outlier features are extracted to highlight the characteristics of abnormal data.Finally,the random forest model is constructed,and the prediction residual is calculated as the measure index of outlier,so as to realize multi-scale mining and mark outliers.The experimental results show that the wireless communication nodes are successfully divided into different clusters after applying this method,and then outliers are accurately identified.With the increase of data volume,the mining execution time is always kept at a low level,which proves that this method has high efficiency in outlier mining.

关 键 词：随机森林 无线通信 多尺度 离群点 挖掘 

分 类 号：TN92[电子电信—通信与信息系统]