基于串行分类算法的不平衡时间序列多分类方法  

Imbalanced Time Series Multi-Classification Method Based on Two-Steps Algorithm

作　　者：陈俐名 黄诗茹 修保新[1] 周鋆 Chen Liming;Huang Shiru;Xiu Baoxin;Zhou Yun(Science and Technology on Information Systems Engineering Laboratory,National University of Defense Technology,Changsha 410073,China)

机构地区：[1]国防科技大学信息系统工程重点实验室

出　　处：《南京师范大学学报（工程技术版）》2019年第3期8-14,共7页Journal of Nanjing Normal University(Engineering and Technology Edition)

基　　金：国家自然科学基金(61703416);湖南省自然科学基金(2018JJ3614)

摘　　要：提出了基于串行分类算法的不平衡时间序列多分类方法,并以“上证50指数”15 min交易数据为例,进行了实验检验与结果分析.结果表明,在多数情况下,串行分类算法比单一算法有更高的准确率、召回率和F1值,可以更有效解决不平衡时间序列多分类问题.This paper proposes a multi-classification method of imbalanced time series based on two-steps classification algorithm,and takes the 15-minutes trading data of“Shanghai Stock Exchange 50 index”as an example to conduct an experimental test and analysis.The results show that,in most cases,the two-steps classification algorithm has a higher accuracy,recall rate and F1 value than the single algorithm,and that it can solve the problem of multiple classification of unbalanced time series more effectively.

关 键 词：不平衡 时间序列 多分类 串行分类算法 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]