基于MCMC方法的区间控制伪装缺失值检测算法  

THE INTERVAL CONTROL DISGUISED MISSING VALUE DETECTION ALGORITHM BASED ON MCMC METHOD

作　　者：史金余[1] 孙禹明 谢兄[1] 刘卫江[1] 滕俊凯 Shi Jinyu;Sun Yuming;Xie Xiong;Liu Weijiang;Teng Junkai(School of Information Science and Technology,Dalian Maritime University,Dalian 116026,Liaoning,China)

机构地区：[1]大连海事大学大学信息科学技术学院,辽宁大连116026

出　　处：《计算机应用与软件》2022年第8期265-272,共8页Computer Applications and Software

基　　金：国家自然科学基金青年科学基金项目(61702074);中央高校基本科研业务费项目(3132019354)。

摘　　要：为减少伪装缺失值对数据分析的严重影响,提出基于MCMC方法的区间控制伪装缺失值检测算法。将MCMC方法和基于统计学的离群点检测算法结合在一起进行检测,选取合适的MCMC方法对参数进行取样,根据采样得到的参数,选取合适的基于统计学的离群点检测算法确定控制区间,在控制区间内对数据进行遍历,判断是否为高频率伪装缺失值。实验表明,该算法在公开数据集上的查全率、查准率、F1-Measure取得了较好的改善。In order to reduce the influence of disguised missing data on data analysis,the interval control disguised missing value detection algorithm based on MCMC method is proposed.The MCMC method and the outlier detection algorithm based on statistics were combined to detect.We selected the appropriate MCMC method to sample the parameters.We selected the appropriate outlier detection algorithm based on statistics to determine control interval according to sampled parameters.The data was traversed in the control interval to determine whether it was a high frequency camouflage missing value.The experimental results show that the recall,precision and F1-Measure of the algorithm on the public data sets are improved well.

关 键 词：伪装缺失值 MCMC方法 控制区间 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]