基于决策树的流数据分类算法综述  

Decision Tree Based Streaming Data Classification Algorithm:A Survey

作　　者：韩成成 增思涛 林强 曹永春[1,2] 满正行[1,2] HAN Cheng-cheng;ZENG Si-taO;LIN Qiang;CAO Yong-chun;MAN Zheng-xing(School of Mathematics and Computer Science,Northwest Minzu University,Lanzhou 730124,China;Key Laboratory of Dynamic Streaming Data Computing and Applications,Lanzhou 730124,China;Key Laboratory of China's Ethnic Languages and Information Technology of Ministry of Education,Northwest Minzu University,Lanzhou 730030,China)

机构地区：[1]西北民族大学数学与计算机科学学院,甘肃兰州730124 [2]西北民族大学动态流数据计算与应用实验室,甘肃兰州730124 [3]西北民族大学中国民族信息技术研究院,甘肃兰州730030

出　　处：《西北民族大学学报（自然科学版）》2020年第2期20-30,共11页Journal of Northwest Minzu University(Natural Science)

基　　金：西北民族大学中央高校基本科研业务费专项资金资助研究生项目(Yxm2020101)。

摘　　要：流数据是一种有别于传统静态数据的新的数据形态,随着时间的推移而不断产生,而且富含变化.流数据分类是数据挖掘的研究分支,用于发现数据中隐含的模式并实现数据的类别划分,通常将每一个类别称作概念.将传统决策树算法引入流数据分类,针对流数据的特征提出特定的分类算法,是流数据分类的一个主要研究分支.为了全面介绍基于决策树的流数据分类算法,首先,简要概述数据挖掘及主要任务、决策树及其主要算法、流数据及其主要特性;然后,按照算法是否考虑概念漂移问题,将现有工作划分为包含概念漂移的流数据分类算法和不含概念漂移的流数据分类算法两大类,分别介绍每一类算法的主要算法流程、优缺点和典型应用;最后,指出基于决策树的流数据分类的进一步研究方向.Data stream is a new data form differing from the traditional static data.The data stream is a sequence of data that are collected in a real-time manner and vary over time.Classifying data stream aims to extract pattern and then classify these data into different categories in the data mining domain,where categories that data stream belong to is often called concepts.Classical decision tree method has been introduced to data stream classification domain,rising a research branch of decision tree based streaming data mining.In order to provide a comprehensive review on decision tree based streaming data classification,in this work,we first present an introduction of data mining and its main tasks,decision tree and its main algorithms,and streaming data and their main properties.Then,existing works on streaming data classification that were developed based on decision tree were detailed by separating these works into two main categories according to if the concept was considered.And last,we pointed out the research challenges and the study direction in the future.

关 键 词：数据挖掘 数据分类 流数据 决策树 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]