基于粗糙集的决策树算法研究及其在汽车售后服务中的应用
|
摘要
随着全球一体化进程的加快,我国汽车行业将面临激烈的竞争。在汽车拥有量持续增长的情况下,汽车售后服务市场也在不断完善。然而,目前客户对汽车售后服务的满意度并不高,究其原因,主要是缺少对客户售后消费行为特征的研究。针对这个问题,论文进行了调查和研究工作,希望能对目前的汽车售后服务市场起到一定的作用。
论文对数据挖掘算法进行了研究,选择粗糙集和决策树作为建模算法。在对相关算法进行研究的基础上,对部分算法进行了改进。
(1)针对传统决策树存在的诸如偏向于选择属性值较多的属性等缺点,通过引入增补权值,提出了WG_ID3算法;
(2)针对传统粗糙集应用于完备信息系统的限制,论文提出一种改进的基于集对分析的可变精度粗糙集的属性约简算法,即SPA_VPRS算法:
(3)针对多变量检测的构造问题所涉及的两个方面,提出一种基于粗糙集的多变量决策树构造算法(即SW_RS_MDT算法)。
通过实验和对比分析,表明论文提出的改进算法与原有传统算法相比,在一定程度上提高了算法的效率和准确性。
作为应用实例,论文构造了基于SW_RS_MDT算法的评估模型,并将该模型应用到汽车售后服务中。通过多次测试评估,提取出影响客户满意度的规则,并为汽车售后服务人员提供服务策略,从而提高客户的满意度,提升汽车售后服务行业的核心竞争能力。
With the acceleration of the process of globalization, the automobile industry of our country will face fierce competition. Under the situation of the car's number being in the continuous growth, the market of automobile after-sales service is also improving constantly. However, nowadays the customer satisfaction is not high. The major reason is the lack of study on the characteristics of consumer's after-sales consumption behavior. Referring to the problem, the article has done a lot of research and investigation, I hope that it can play a certain role in the current automobile after-sales service market.
The article researches the Data Mining algorithm, and it selected Rough Set and Decision Tree as modeling algorithm. On the basis of researching relative algorithm, the article made some improvement on some parts of algorithm.
(1) Referring to the defects of traditional Decision Tree, such as it inclines to choose the attributes who have more value, the article comes up with WGID3 algorithm by introducing the addition weight.
(2) Referring to the restriction that the traditional Rough Set is applied to complete system, the article introduces an attribute reduction algorithm which is based on Set Pair Analysis and Variable Precision Rough Set, that is SPA_VPRS algorithm.
(3) Referring to the two aspects of the construction problem for multivariate testing, the article introduces a multivariate decision tree construction algorithm on the basis of Rough Set, that is SW_RS_MDT algorithm.
Through experiments and comparative analysis, the article proves that, to a certain extent, these improved algorithm are more exact and more effective than traditional algorithm.
As examples of applications, the article constructed evaluation model which is based on SW_RS_MDT algorithm and the model is applied to automobile after-sales service. Through testing and evaluation many times, the extracted rules which affects customer satisfaction degree can provide service strategy for the people who participate in automobile after-sales service, it can improve customer satisfaction degree and enhance the core competitiveness of automobile after-sales service industry.
论文对数据挖掘算法进行了研究,选择粗糙集和决策树作为建模算法。在对相关算法进行研究的基础上,对部分算法进行了改进。
(1)针对传统决策树存在的诸如偏向于选择属性值较多的属性等缺点,通过引入增补权值,提出了WG_ID3算法;
(2)针对传统粗糙集应用于完备信息系统的限制,论文提出一种改进的基于集对分析的可变精度粗糙集的属性约简算法,即SPA_VPRS算法:
(3)针对多变量检测的构造问题所涉及的两个方面,提出一种基于粗糙集的多变量决策树构造算法(即SW_RS_MDT算法)。
通过实验和对比分析,表明论文提出的改进算法与原有传统算法相比,在一定程度上提高了算法的效率和准确性。
作为应用实例,论文构造了基于SW_RS_MDT算法的评估模型,并将该模型应用到汽车售后服务中。通过多次测试评估,提取出影响客户满意度的规则,并为汽车售后服务人员提供服务策略,从而提高客户的满意度,提升汽车售后服务行业的核心竞争能力。
With the acceleration of the process of globalization, the automobile industry of our country will face fierce competition. Under the situation of the car's number being in the continuous growth, the market of automobile after-sales service is also improving constantly. However, nowadays the customer satisfaction is not high. The major reason is the lack of study on the characteristics of consumer's after-sales consumption behavior. Referring to the problem, the article has done a lot of research and investigation, I hope that it can play a certain role in the current automobile after-sales service market.
The article researches the Data Mining algorithm, and it selected Rough Set and Decision Tree as modeling algorithm. On the basis of researching relative algorithm, the article made some improvement on some parts of algorithm.
(1) Referring to the defects of traditional Decision Tree, such as it inclines to choose the attributes who have more value, the article comes up with WGID3 algorithm by introducing the addition weight.
(2) Referring to the restriction that the traditional Rough Set is applied to complete system, the article introduces an attribute reduction algorithm which is based on Set Pair Analysis and Variable Precision Rough Set, that is SPA_VPRS algorithm.
(3) Referring to the two aspects of the construction problem for multivariate testing, the article introduces a multivariate decision tree construction algorithm on the basis of Rough Set, that is SW_RS_MDT algorithm.
Through experiments and comparative analysis, the article proves that, to a certain extent, these improved algorithm are more exact and more effective than traditional algorithm.
As examples of applications, the article constructed evaluation model which is based on SW_RS_MDT algorithm and the model is applied to automobile after-sales service. Through testing and evaluation many times, the extracted rules which affects customer satisfaction degree can provide service strategy for the people who participate in automobile after-sales service, it can improve customer satisfaction degree and enhance the core competitiveness of automobile after-sales service industry.
引文
[1]欣华.整车利润低于制造业平均水平.市场报.2006年2月27日[第九版].
[2]张莹,梅强.我国汽车销售4S店的现状及原因分析[J].企业活力,2006[5]:28-29.
[3]郑刚.中国乘用车售后服务市场消费者五大行为特征.汽车与配件,产业经济频道,2007,33[8]:34-35.
[4]Mehmed Kantardzic.《数据挖掘-概念、模型、方法和算法》[M].北京:清华大学出版社,2003.
[5]Quinlan J R.Induction of decision tree[J].Machine Learning,1986,[1]:81-106.
[6]Quinlan J R.C4.5:Programs for Machine Learning.Morgan Kaufman,1993.
[7]Boz O.Extracting decision trees from trained neural networks[C].Edmonton,Alberta,Canada:Proc of the Eighth ACM SIGKDD International Conference on Knowledge Discovery and Data Mining,2002.
[8]Cristina Olaru,Louis Wehenkel.A complete fuzzy decision tree technique.Fuzzy Sets and Systems.2003:138[2].
[9]Cantu Paz E,Kamath C.Inducing oblique decision trees with evolutionary Algorithms[J].IEEE Transactions on Evolutionary Computation,2003,7[1]:54-68.
[10]潘珩.决策树技术在汽车销售中的应用研究.重庆工商大学学报[自然科学版].2006[4]:418-423.
[11]房华蓉,张毅.微软决策树在汽车销售中的应用.电脑知识与技术[学术交流].2007[4]:1045-1046,1068.
[12]曾黄麟.粗集理论及其应用[M].重庆:重庆大学出版社.1998.
[13]Naisbitt J.Megatrends:Ten new directions transforming our lives[M].New York:Warner Books,1982.16-17.
[14]Chen M S,Han J,Yu P S.Data Mining:An Overview from a Database Perspective[J].IEEE Transaction on Knowledge and Data Engineering,1996,18[6]:1-14.
[15]Fayyad U.Piatetsky-Shapiro,Smith,Uthurusamy.Advances in Knowledge Discovery and Data Mining,America[M],AAA/MIT Press,1996,7:1-35.
[16]谢丹夏,李晓东.数据挖掘技术在Web上的应用及其工具设计.计算机 应用,2001[2]:42-44.
[17]G.Piatetsky-Shapiro,Knowledge Discovery in Databases,AAAI/MIT Press,1991.
[18]石杰楠.数据挖掘研究综述[J].航天制造技术,2005[4]:27-31.
[19]张倩.数据挖掘技术综述[J].甘肃科技,2005[7]:92-93.
[20]曹坤.决策树遗传算法融合模型及其在电信业客户流失分析中的应用研究[D].景德镇.景德镇陶瓷学院,2005:1-9.
[21]范明,孟小峰译.数据挖掘-概念与技术[M].北京:机械工业出版社.2001.
[22]石纯一,黄昌宁等.人工智能原理[M].北京:清华大学出版社.1993.
[23]刘红岩,陈剑等.数据挖掘中的数据分类算法综述[J].清华大学学报[自然科学版].2002,42[6]:727-730.
[24]Tom M.Mitchell著,[美]卡内基梅隆大学,机器学习[M],2003年1月第一版,曾华军,张银奎等译.北京:机械工业出版社,2003.
[25]蔡自兴,徐光佑.人工智能及其应用[M].北京:清华大学出版社.1996.
[26]Jiawei Han,Micheline Kamber.数据挖掘概念与技术[M].北京:机械工业出版社,2001.
[27]胡可云.基于概念格和粗糙集的数据挖掘方法研究.[D].北京.清华大学.2001.
[28]T.M.Mitchell.Machine Learning.New York:McGraw-Hill.1997.
[29]Pawlak Z.Rough sets[J].International Journal of Computer and Information Science,1982,11:341-356.
[30]史忠植.知识发现[M].北京:清华大学出版社,2002.
[31]Sreerama,K.Murthy.Automatic Construction of Decision Trees from Data:A Multi-Disciplinary Survey.Data Mining and Knowledge Discovery.1998,2:345-389.
[32]Hunt E B,J Marin,P T Stone.1996.Experiment in Induction.Academic Press.
[33]L.Breiman,J.H.Friedman,R.A.Olshen,and C.J.Stone.Classification and Regression Trees.Wadsworth,Belmont,1984.
[34]M.Mehta,R.AgrawaI.SLIQ:A fast scalable classifier for data mining.Int.Conf.EDBT'96 28.
[35]J.Shafer,R.Agrawal,and M.Mehta.SPRINT:A Scalable parallel classifier for data mining.In Proc.1996 Int.Conf.VLDB'96,p544-555.Bombay,India,Sept.1996.
[36]Pawlak Z.Rough sets[J].International Journal of Computer and Information Science,1982,11:341-356.
[37]韩少锋,陈立潮.浅谈粗糙集理论及其应用进展[J].山西电子技术.2006.1:92-93.
[38]王珏.粗糙集理论及其应用研究[D].西安:西安电子科技大学,2005.
[39]关晓蔷,刘煜伟.一种基于粗糙集的决策树构造方法[J].科学情报开发与经济.2006,16[13]:136-137.
[40]王艳兵.基于决策树和粗糙集的分类方法研究[D].济南:山东大学,2006.
[41]张冬艳.基于粗糙集合理论的决策树构造算法研究[D].合肥:合肥工业大学,2006.
[42]王国胤.Rough集理论与知识获取[M].西安:西安交通大学出版社.2001.
[43]张文修,梁怡等.信息系统与知识发现[M].北京:科学出版社,2003.
[44]张文修,吴志伟.粗糙集理论与发现[M].北京:科学出版社,2001.
[45]韩祯祥,张琦.粗糙集理论及其应用综述[J].控制理论与应用,1999,16[2]:153-157.
[46]Zearko W.Variable Precision Rough Set Mode[J].Journal of Computer and System Sciences.1993,46:39-59.
[47]Wong S KM,Ziarko W.On optimal decision rules in decision tables.Bulletin of Polish Academy of Sciences,1985,33:693-696.
[48]Pawlak Z,Slowinski K.Rough Set Approach to Multi-attribute Decision Analysis.European Journal of Operational Research,1994,72:443-459.
[49]A.Skowron,C.Rauszer,The discernibility matrices and functions in information systems,in:E.Stowiflski[Ed.],Intelligent Decision Support Handbook of Applications and Advances of the Rough Sets Theory,Kluwer Academic Publishers,Dodrecht,1992:331-362.
[50]J.Z Dong,N.Zhong,and S.Ohsuga.Using Rough Sets with Heuristics to Feature Selection[J].Journal of Intelligent information Systems,2001,16:199-214.
[51]李玉榕,乔斌等.基于熵的粗糙集属性约简算法.电路与系统学报.2002[3]:8-12.
[52]Ziarko W,Cercone N,Hu X.Rule Discovery from Database with Decision Matrices,9~(th)Int.Symposium on Foundation of Intelligent Systems,1996:653-662.
[53]Marzena Kryszkiewicz Rough Set Approach to Incomplete Information system.Information Sciences,1998,112:39-49.
[54]Therasa Beaubouef,Frederick E.Petry,Gurdial Arora,Information-theoretic Measures of Uncertainty for Rough Sets and Rough Relational Databases,Journal of Information Sciences,1998,109:185-195.
[55]苗夺谦,王珏.基于粗糙集的多变量决策树构造方法[J].软件学报,1997,8[6]:425-431.
[56]Kohavi R,Frasca B.Userful feature subsets and rough set reducts[A].Proc on RSSC'94[C].1994.
[57]罗秋瑾,马锐.基于粗糙集和熵的多变量决策树的构造算法[J].计算机应用.1997,27[7]:1708-1710.
[58]王国胤.Rough集理论在不完备信息系统中的扩充[J].计算机研究与发展,2002,39[10]:1238-1243.
[59]赵克勤.集对分析及其初步应用[M].杭州:浙江科学出版社,2000.
[60]郭明,郑惠莉.用数据挖掘法分析电信客户流失[J].现代通信.2005[3]:7-9.
[61]方坤.移动通信经营分析系统的构建与客户流失分析[D].南京:南京航空航天大学.2004.
[2]张莹,梅强.我国汽车销售4S店的现状及原因分析[J].企业活力,2006[5]:28-29.
[3]郑刚.中国乘用车售后服务市场消费者五大行为特征.汽车与配件,产业经济频道,2007,33[8]:34-35.
[4]Mehmed Kantardzic.《数据挖掘-概念、模型、方法和算法》[M].北京:清华大学出版社,2003.
[5]Quinlan J R.Induction of decision tree[J].Machine Learning,1986,[1]:81-106.
[6]Quinlan J R.C4.5:Programs for Machine Learning.Morgan Kaufman,1993.
[7]Boz O.Extracting decision trees from trained neural networks[C].Edmonton,Alberta,Canada:Proc of the Eighth ACM SIGKDD International Conference on Knowledge Discovery and Data Mining,2002.
[8]Cristina Olaru,Louis Wehenkel.A complete fuzzy decision tree technique.Fuzzy Sets and Systems.2003:138[2].
[9]Cantu Paz E,Kamath C.Inducing oblique decision trees with evolutionary Algorithms[J].IEEE Transactions on Evolutionary Computation,2003,7[1]:54-68.
[10]潘珩.决策树技术在汽车销售中的应用研究.重庆工商大学学报[自然科学版].2006[4]:418-423.
[11]房华蓉,张毅.微软决策树在汽车销售中的应用.电脑知识与技术[学术交流].2007[4]:1045-1046,1068.
[12]曾黄麟.粗集理论及其应用[M].重庆:重庆大学出版社.1998.
[13]Naisbitt J.Megatrends:Ten new directions transforming our lives[M].New York:Warner Books,1982.16-17.
[14]Chen M S,Han J,Yu P S.Data Mining:An Overview from a Database Perspective[J].IEEE Transaction on Knowledge and Data Engineering,1996,18[6]:1-14.
[15]Fayyad U.Piatetsky-Shapiro,Smith,Uthurusamy.Advances in Knowledge Discovery and Data Mining,America[M],AAA/MIT Press,1996,7:1-35.
[16]谢丹夏,李晓东.数据挖掘技术在Web上的应用及其工具设计.计算机 应用,2001[2]:42-44.
[17]G.Piatetsky-Shapiro,Knowledge Discovery in Databases,AAAI/MIT Press,1991.
[18]石杰楠.数据挖掘研究综述[J].航天制造技术,2005[4]:27-31.
[19]张倩.数据挖掘技术综述[J].甘肃科技,2005[7]:92-93.
[20]曹坤.决策树遗传算法融合模型及其在电信业客户流失分析中的应用研究[D].景德镇.景德镇陶瓷学院,2005:1-9.
[21]范明,孟小峰译.数据挖掘-概念与技术[M].北京:机械工业出版社.2001.
[22]石纯一,黄昌宁等.人工智能原理[M].北京:清华大学出版社.1993.
[23]刘红岩,陈剑等.数据挖掘中的数据分类算法综述[J].清华大学学报[自然科学版].2002,42[6]:727-730.
[24]Tom M.Mitchell著,[美]卡内基梅隆大学,机器学习[M],2003年1月第一版,曾华军,张银奎等译.北京:机械工业出版社,2003.
[25]蔡自兴,徐光佑.人工智能及其应用[M].北京:清华大学出版社.1996.
[26]Jiawei Han,Micheline Kamber.数据挖掘概念与技术[M].北京:机械工业出版社,2001.
[27]胡可云.基于概念格和粗糙集的数据挖掘方法研究.[D].北京.清华大学.2001.
[28]T.M.Mitchell.Machine Learning.New York:McGraw-Hill.1997.
[29]Pawlak Z.Rough sets[J].International Journal of Computer and Information Science,1982,11:341-356.
[30]史忠植.知识发现[M].北京:清华大学出版社,2002.
[31]Sreerama,K.Murthy.Automatic Construction of Decision Trees from Data:A Multi-Disciplinary Survey.Data Mining and Knowledge Discovery.1998,2:345-389.
[32]Hunt E B,J Marin,P T Stone.1996.Experiment in Induction.Academic Press.
[33]L.Breiman,J.H.Friedman,R.A.Olshen,and C.J.Stone.Classification and Regression Trees.Wadsworth,Belmont,1984.
[34]M.Mehta,R.AgrawaI.SLIQ:A fast scalable classifier for data mining.Int.Conf.EDBT'96 28.
[35]J.Shafer,R.Agrawal,and M.Mehta.SPRINT:A Scalable parallel classifier for data mining.In Proc.1996 Int.Conf.VLDB'96,p544-555.Bombay,India,Sept.1996.
[36]Pawlak Z.Rough sets[J].International Journal of Computer and Information Science,1982,11:341-356.
[37]韩少锋,陈立潮.浅谈粗糙集理论及其应用进展[J].山西电子技术.2006.1:92-93.
[38]王珏.粗糙集理论及其应用研究[D].西安:西安电子科技大学,2005.
[39]关晓蔷,刘煜伟.一种基于粗糙集的决策树构造方法[J].科学情报开发与经济.2006,16[13]:136-137.
[40]王艳兵.基于决策树和粗糙集的分类方法研究[D].济南:山东大学,2006.
[41]张冬艳.基于粗糙集合理论的决策树构造算法研究[D].合肥:合肥工业大学,2006.
[42]王国胤.Rough集理论与知识获取[M].西安:西安交通大学出版社.2001.
[43]张文修,梁怡等.信息系统与知识发现[M].北京:科学出版社,2003.
[44]张文修,吴志伟.粗糙集理论与发现[M].北京:科学出版社,2001.
[45]韩祯祥,张琦.粗糙集理论及其应用综述[J].控制理论与应用,1999,16[2]:153-157.
[46]Zearko W.Variable Precision Rough Set Mode[J].Journal of Computer and System Sciences.1993,46:39-59.
[47]Wong S KM,Ziarko W.On optimal decision rules in decision tables.Bulletin of Polish Academy of Sciences,1985,33:693-696.
[48]Pawlak Z,Slowinski K.Rough Set Approach to Multi-attribute Decision Analysis.European Journal of Operational Research,1994,72:443-459.
[49]A.Skowron,C.Rauszer,The discernibility matrices and functions in information systems,in:E.Stowiflski[Ed.],Intelligent Decision Support Handbook of Applications and Advances of the Rough Sets Theory,Kluwer Academic Publishers,Dodrecht,1992:331-362.
[50]J.Z Dong,N.Zhong,and S.Ohsuga.Using Rough Sets with Heuristics to Feature Selection[J].Journal of Intelligent information Systems,2001,16:199-214.
[51]李玉榕,乔斌等.基于熵的粗糙集属性约简算法.电路与系统学报.2002[3]:8-12.
[52]Ziarko W,Cercone N,Hu X.Rule Discovery from Database with Decision Matrices,9~(th)Int.Symposium on Foundation of Intelligent Systems,1996:653-662.
[53]Marzena Kryszkiewicz Rough Set Approach to Incomplete Information system.Information Sciences,1998,112:39-49.
[54]Therasa Beaubouef,Frederick E.Petry,Gurdial Arora,Information-theoretic Measures of Uncertainty for Rough Sets and Rough Relational Databases,Journal of Information Sciences,1998,109:185-195.
[55]苗夺谦,王珏.基于粗糙集的多变量决策树构造方法[J].软件学报,1997,8[6]:425-431.
[56]Kohavi R,Frasca B.Userful feature subsets and rough set reducts[A].Proc on RSSC'94[C].1994.
[57]罗秋瑾,马锐.基于粗糙集和熵的多变量决策树的构造算法[J].计算机应用.1997,27[7]:1708-1710.
[58]王国胤.Rough集理论在不完备信息系统中的扩充[J].计算机研究与发展,2002,39[10]:1238-1243.
[59]赵克勤.集对分析及其初步应用[M].杭州:浙江科学出版社,2000.
[60]郭明,郑惠莉.用数据挖掘法分析电信客户流失[J].现代通信.2005[3]:7-9.
[61]方坤.移动通信经营分析系统的构建与客户流失分析[D].南京:南京航空航天大学.2004.