考虑分类代价的用户换手机的分类器研究——基于数据平衡性视角

管理评论 ›› 2021, Vol. 33 ›› Issue (9): 177-186.

考虑分类代价的用户换手机的分类器研究——基于数据平衡性视角

王超发¹, 王树斌², 王成军³

1. 西安电子科技大学经济与管理学院, 西安 710126;
2. 西安邮电大学经济与管理学院, 西安 710061;
3. 西安建筑科技大学管理学院, 西安 710055

收稿日期:2018-06-08 出版日期:2021-09-28 发布日期:2021-10-09
通讯作者: 王超发(通讯作者),西安电子科技大学经济与管理学院副教授,博士
作者简介:王树斌,西安邮电大学经济与管理学院讲师,博士;王成军,西安建筑科技大学管理学院教授,博士生导师,博士。
基金资助:
国家自然科学基金面上项目（71872141；51774228）。

Study on the Classifier for Mobile Phone Users Considering the Classification Cost——Based on the Perspective of Data Balance

Wang Chaofa¹, Wang Shubin², Wang Chengjun³

1. School of Economics and Management, Xidian University, Xi'an 710126;
2. School of Economics and Management, Xi'an University of Posts and Telecommunications, Xi'an 710061;
3. School of Management, Xi'an University of Architecture and Technology, Xi'an 710055

Received:2018-06-08 Online:2021-09-28 Published:2021-10-09

摘要/Abstract

摘要： 传统分类算法在处理非平衡数据时不能有效提高用户的分类效果。本文基于中国移动通信集团公司某分公司的用户数据，从数据平衡性视角出发，对判别用户是否换手机的分类器最优阈值、预期风险和分类代价之间的关系进行了实证分析。研究发现：以平衡数据集为样本对用户换手机进行分类得到的换机准确率高于原始数据集对应的换机准确率；预期风险（同一分类代价比）随着分类器阈值的增大表现出先增大后减小的趋势；对用户是否换手机的分类结果受数据平衡性和分类代价比的双重影响。研究结论能够为手机销售商和制造商完善管理方法提供决策依据。

关键词: 非平衡数据, 分类器, 分类准确率, 分类代价

Abstract: Traditional classification algorithms cannot effectively improve the classification effect of users when dealing with unbalanced data. Based on the user information data from a branch of China Mobile Communications Corporation, this study makes, from the data balance perspective, an empirical analysis on mobile phone classifier, the optimal threshold, expected risk, and the relationship between the classification of the cost. Research findings:the accuracy of mobile phone exchange is higher than that of the original data set; With the increase of the threshold value of the classifier, same classification cost ratio, the expected risk appears to increase first and then decrease; the classification result of whether the user changes the mobile phone or not is influenced by the balance of data and the set of classification cost ratio. These conclusions can provide a decision-making basis for mobile phone sellers and manufacturers to improve their management methods.

Key words: non-equilibrium data, classifier, classification accuracy, classification cost

王超发, 王树斌, 王成军. 考虑分类代价的用户换手机的分类器研究——基于数据平衡性视角[J]. 管理评论, 2021, 33(9): 177-186.

Wang Chaofa, Wang Shubin, Wang Chengjun. Study on the Classifier for Mobile Phone Users Considering the Classification Cost——Based on the Perspective of Data Balance[J]. Management Review, 2021, 33(9): 177-186.

参考文献

[1] Zong W., Huang G. B., Chen Y. Weighted Extreme Learning Machine for Imbalance Learning[J]. Eeuro-computing, 2013,101(3):229-242
[2] Alberto F., María J. J., Francisco H. Hierarchical Fuzzy Rule Based Classification Systems with Genetic Rule Selection for Imbalanced Data-sets[J]. International Journal of Approximate Reasoning, 2009,50(8):561-577
[3] Hungyuan C., Chihhsiang H., Chechang H. Support Vector Machines Using Bayesian-based Approach in the Issue of Unbalanced Classifications[J]. Expert Systems with Applications, 2011,38(6):11447-11452
[4] Folino G., Pisani F. S. Evolving Meta-ensemble of Classifiers for Handling Incomplete and Unbalanced Datasets in the Cyber Security Domain[J]. Applied Soft Computing, 2016,47(2):179-190
[5] Miriam R. S., Lee L. L. Pruning Methods to MLP Neural Networks Considering Proportional Apparent Error Rate for Classification Problems with Unbalanced Data[J]. Measurement, 2014,56(5):88-94
[6] Silvia C., Valentina C., Marco V. A Method for Resampling Imbalanced Datasets Inbinary Classification Tasks for Real-world Problems[J]. Neurocomputing, 2014,135(8):32-41
[7] Ganesh S. G., Vadlamani R. A Novel Hybrid Under Sampling Method Formining Unbalanced Datasets in Banking and Insurance[J]. Engineering Applications of Artificial Intelligence, 2015,37(10):368-377
[8] Ammar N., Hadi J., Mariette A. Kermin SVM for Imbalanced Datasets with a Case Study on Arabic Comics Classification[J]. Engineering Applications of Artificial Intelligence, 2017,59(4):59-169
[9] Lixiang D., Mengyun X., Tangbo B., et al. A New Support Vector Data Description Method for Machinery Fault Diagnosis with Unbalanced Datasets[J]. Expert Systems with Applications, 2016,64(11):239-246
[10] Fan X. N., Tang K., Weise T. Margin-Based Over-Sampling Method for Learning from Imbalanced Datasets[C]. Springer Berlin Heidelberg, 2011,66(35):309-320
[11] Tao W., Zhenxing Q., Shichao Z., et al. Cost-sensitive Classification with Inadequate Labeled Data[J]. Information Systems, 2012,37(5):508-516
[12] Chengjian X., Wenxuan Z., Wenzhi H., et al. The Situation of Waste Mobile Phone Management in Developed Countries and Development Status in China[J]. Waste Management, 2016,58(9):341-347
[13] Bukelwa N., Stephen V. F. Information Security Behaviour Profiling Framework (ISBPF) for Student Mobile Phone Users[J]. Science Direct, 2015,53(12):132-142
[14] Teresa G., Teodosio P., Covadonga G., et al. Consumer Complaint Behaviour in Telecommunications:The Case of Mobile Phone Users in Spain[J]. Telecommunications Policy, 2016,40(5):804-820
[15] Anna S., József M., Pirkko W. An Attitude-based Latent Class Segmentation Analysis of Mobile Phone Users[J]. Telematics and Informatics, 2014,31(1):209-219
[16] Victoria K., Sharmistha B. S., Raghav R. Extent of Private Information Disclosure on Online Social Network s:An Exploration of Facebook Mobile Phone Users[J]. Computers in Human Behavior, 2013,29(8):2722-2729
[17] Veronika K., Akos P., Adam M. Mobile Attachment:Separation from the Mobile Phone Induces Physiological and Behavioural Stress and Attentional Bias to Separation-related Stimuli[J]. Computers in Human Behavior, 2017,71(3):228-239
[18] Liua C. C., Chen C. H. Using Q Methodology to Explore User's Value Types on Mobile Phone Service Websites[J]. Expert Systems with Applications, 2013,40(2):5276-5283
[19] Feng L., Shaoqiong Z., Yang L. How Many, How Often, and How New? A Multivariate Profiling of Mobile App Users[J]. Journal of Retailing and Consumer Services, 2017,38(12):71-80
[20] Richard N., Frank N. A Real Generalization of Discrete AdaBoost[J]. Artificial Intelligence, 2007,171(1):25-41
[21] 胡海青,张琅,张道宏. 供应链金融视角下的中小企业信用风险评估研究——基于SVM与BP神经网络的比较研究[J]. 管理评论, 2012,24(11):70-80
[22] 郭亚伟,白治江. 基于混合重采样的非平衡数据SVM训练方法[J]. 人工智能, 2016,35(12):52-58
[23] Jungeun K., Keunho C., Gunwoo K., et al. Classification Cost:An Empirical Comparison among Traditional Classifier, Cost-Sensitive Classifier, and MetaCost[J]. Expert Systems with Applications, 2012,39(4):4013-4019
[24] 李秋洁,赵亚琴,顾洲. 代价敏感学习中的损失函数设计[J]. 控制理论与应用, 2015,32(5):689-694
[25] Lu S. H., Chiang D. A., Keh H. C., et al. Chinese Text Classification by the Naïve Bayes Classifier and the Associative Classifier with Multiple Confidence Threshold Values[J]. Knowledge-Based Systems, 2010,23(6):598-604
[26] 郑恩辉. 基于支持向量机的代价敏感数据挖掘研究与应用[D]. 浙江大学博士学位论文, 2006
[27] 贾国柱,刘圣国,王剑磊,等. 基于支持向量机的建筑企业循环经济评价研究[J]. 管理评论, 2013,25(5):11-18
[28] Dongliang W., Lili T., Yichuan Z. Smoothed Empirical Likelihood for the Youden Index[J]. Computational Statistics & Data Analysis, 2017,115(9):1-10