基于 DTW_M 的时序邻域特征选择算法

打开文本图片集

关键词：特征选择；高维时序数据； DTWM 度量；马氏距离；邻域粗糙集中图分类号：TP391 文献标志码：A 文章编号：1001-3695（2026）01-020-0170-08doi：10.19734/j. issn.1001-3695.2026.05.0157

Feature selection of time-series neighborhood based on DTWy metric

Yang Xuan1，²，Wang Xiaowan³†，Hu Lingzhi 1 ，Wu Di1 （1.ScholofBasicMedicalSciences，Shaanxi Universityofhinese Medicine，XianyangShanxi72O46，China；2.ScholofSciences， Chang'an University，Xi’an710064，China；3.School ofSofware，Tsinghua University，Beijing 184，China）

Abstract：High-dimensionaltimeseriesdataexistwidelyinreallife.Theyoften havedecisionatributesandvaryintime length.Thesecharacteristicsrender existingneighborhoodroughsetfeatureselectionalgorithms inaplicableorreducingtheir classficationperformance.Thus，thispaperproposedafeatureselectionmethodforhigh-dimensional timeseriesdatabasedon metrics.Firstly，it introduced the Mahalanobis distance and defined thedynamic time warping of Mahalanobis（DTW M ）to measurethesimilaritybetweenatrbutes.Then，italsodefinedatimeseriesdecisioninformationsystemtostorenon-equallengthhigh-dimensionaltimeseriesdata.Italsoproposedtimeseriesneighborhoodrelationshipandatimeseriesneighborhood rough set model based on DTW M distance measurement.Finally，it defined internal and external importance，and presented the atributedependencyservedasakeyindicatorforscreeningandselectingatributes.Therebythispaperputforwarda feature selection method for high-dimensional time series data based on DTWM measurement. Experiments on five public datasetsverify the method has an average improvement of 14.2% and 21.7% in classification accuracy. These results fully validate the effectiveness and superiority of the proposed method.

Key words： feature selection;high-dimensional time-series data; DTWM metric;Mahalanobis distance；neighborhood rough set

0 引言

随着信息技术的飞速发展，海量高维特征的数据呈爆炸式增长，同时数据的相关性和不确定性也随之而来。（剩余17526字）