基于改进LOF的高维数据异常检测方法

Anomaly detection for high-dimensional data based on improved LOF

作为数据挖掘的核心问题之一,检测离群点或异常值是及时发现故障和隐患问题的重要判断依据。随着物联网设备量的持续增长,传统的单维异常检测算法已经难以满足日益复杂的大数据应用场景。对多维、庞大的数据流进行异常检测时,容易发生检测速度慢和研判准确度下降的问题。本文提出了一个基于高维数据的改进LOF异常检测算法,以提高检测速度和检测精度。同时构建了一个面向海量监控指标数据的流式处理框架,保障异常检测的正常运行。实验结果表明,改进后的算法在准确率和计算效率上有明显提升。

As one of the core issues of data mining,anomaly detection is an important basis for judging faults and hidden dangers in time.With the development of computer technology and the continuous growth of data volume,traditional single-dimensional anomaly detection algorithms have been unable to meet increasingly complex big data application scenarios,which might lead to the problem of slow detection speed and reduce accuracy of research and judgment.In this paper,an improved LOF anomaly detection algorithm based on high-dimensional data is proposed to improve the detection speed and detection accuracy.At the same time,a streaming processing framework for massive monitoring index data is constructed to ensure the normal operation of anomaly detection.The final experimental results show that the improved algorithm has a signifi cant improvement in accuracy and computational effi ciency.