机器学习在食品安全风险预警及抽检方案制订中的应用研究

管理评论 ›› 2022, Vol. 34 ›› Issue (11): 315-323.

机器学习在食品安全风险预警及抽检方案制订中的应用研究

杨鸿雁¹, 田英杰^1,2

1. 中国科学院大学经济与管理学院, 北京 100190;
2. 中国科学院虚拟经济与数据科学研究中心, 北京 100190

收稿日期:2020-06-02 出版日期:2022-11-28 发布日期:2022-12-30
通讯作者: 田英杰(通讯作者),中国科学院大学经济与管理学院,中国科学院虚拟经济与数据科学研究中心研究员,博士生导师,博士。
作者简介:杨鸿雁,中国科学院大学经济与管理学院博士研究生。

Application Research of Machine Learning in Food Safety Risk Early Warning and Sampling Inspection Program

Yang Hongyan¹, Tian Yingjie^1,2

1. School of Economics and Management, University of Chinese Academy of Sciences, Beijing 100190;
2. CAS Research Center on Fictitious Economy & Data Science, Beijing 100190

Received:2020-06-02 Online:2022-11-28 Published:2022-12-30

摘要/Abstract

摘要： “问题导向”的食品安全监督抽检方案对加强食品安全风险管理具有极重要作用。现有关于食品抽检数据的研究较少考虑到不合格样本极少的不均衡特性。本文基于新疆2015—2017年的食品监督抽检数据,将不均衡问题解决思路引入对食品安全风险预警模型的构建研究中,分别构建多种采样方法、代价敏感方法与SVM、随机森林的组合算法模型。经分析比较,在解决食品抽检不均衡数据的问题方面,采样方法逊于代价敏感方法。关于风险预警模型的构建,代价敏感+SVM的组合性能虽稍优于代价敏感+随机森林组合,但前者的运行时间远多于后者。实证结果表明,基于不均衡数据解决思路构建的组合算法模型能够有效提升食品安全风险预警效果,为制订科学合理的食品安全监督抽检方案、提升食品安全风险管理效能提供决策支持。

关键词: 监督抽检, 不均衡问题, 组合算法模型, 风险预警

Abstract: The problem-oriented sampling inspection program of food safety supervision plays an extremely important role in strengthening food safety risk management. Existing researches on food sampling data rarely consider the unbalanced characteristics of unqualified samples. Based on the food inspection and sampling data of Xinjiang from 2015 to 2017, this paper introduces the solution of unbalanced problems into the research on the construction of early warning models of food safety risks, and constructs a combined algorithm model of multiple sampling methods, cost-sensitive methods and SVM, and random forest . Our analysis and comparison show that sampling methods are inferior to cost-sensitive method in solving the problem of unbalanced data in food sampling. Regarding the construction of the risk warning model, although the cost-sensitive + SVM combination performance is slightly better than the cost-sensitive + random forest combination, the running time of the former is much longer than the latter. The empirical results show that the combined algorithm model based on imbalanced data solution can effectively improve the early warning effect of food safety risks, and provide decision support for formulating scientific and reasonable sampling inspection program of food safety supervision and improving the effectiveness of food safety risk management.

Key words: supervision and sampling inspection, imbalance problem, combined algorithm model, risk warning

杨鸿雁, 田英杰. 机器学习在食品安全风险预警及抽检方案制订中的应用研究[J]. 管理评论, 2022, 34(11): 315-323.

Yang Hongyan, Tian Yingjie. Application Research of Machine Learning in Food Safety Risk Early Warning and Sampling Inspection Program[J]. Management Review, 2022, 34(11): 315-323.

参考文献

[1] 王星云,左敏,肖克晶,等. 基于BP神经网络的食品安全抽检数据挖掘[J]. 食品科学技术学报, 2016,34(6):85-90
[2] 游清顺,王建新,张秀宇,等. 基于支持向量机的食品安全抽检数据分析方法[J]. 软件工程, 2019,22(2):29-31
[3] Liu Y., Li X., Wang J., et al. Pattern Discovery from Big Data of Food Sampling Inspections Based on Extreme Learning Machine[C]. International Conference on Research and Practical Issues of Enterprise Information Systems, Springer, Cham, 2017
[4] Zhang R., Zhou L., Zuo M., et al. Prediction of Dairy Product Quality Risk Based on Extreme Learning Machine[C]. International Conference on Data Science & Business Analytics, IEEE Computer Society, 2018
[5] Khosa I., Pasero E., Defect Detection in Food Ingredients Using Multilayer Perceptron Neural Network[C]. 2014 World Symposium on Computer Applications & Research (WSCAR), 2014
[6] 陈恺. 基于数据驱动的食品安全预警分析方法与应用[D]. 北京化工大学博士学位论文, 2015
[7] 陶新民,郝思媛,张冬雪,等. 不均衡数据分类算法的综述[J]. 重庆邮电大学学报(自然科学版), 2013,25(1):101-110
[8] Chawla N. V., Bowyer K. W., Hall L. O., et al. SMOTE:Synthetic Minority Over-sampling Technique[J]. Journal of Artificial Intelligence Research, 2002,16(1):321-357
[9] 覃朗,朱建军,衣柏衡,等. 非均衡数据下基于信息增益的SMOTE改进SVM模型研究[J]. 中国管理科学, 2016,24(S1):128-136
[10] 薛薇. 非平衡数据集的改进SMOTE再抽样算法[J]. 统计研究, 2012,29(6):95-98
[11] 马景义,胡迪,孙枭枭. 不平衡数据处理的新方法——基于样本相似度的少数类合成法[J]. 数理统计与管理, 2015,34(5):809-820
[12] 衣柏衡,朱建军,李杰. 基于改进SMOTE的小额贷款公司客户信用风险非均衡SVM分类[J]. 中国管理科学, 2016,24(3):24-30
[13] Han X., Cui R., Lan Y., et al. A Gaussian Mixture Model Based Combined Resampling Algorithm for Classification of Imbalanced Credit Data Sets[J]. International Journal of Machine Learning and Cybernetics, 2019,10(12):3687-3699
[14] 夏利宇,何晓群. 基于重抽样法处理不平衡问题的信用评级模型[J]. 管理评论, 2020,32(3):75-84
[15] 熊冰妍,王国胤,邓维斌. 基于样本权重的不平衡数据欠抽样方法[J]. 计算机研究与发展, 2016,53(11):2613-2622
[16] 牛壮,李凤莲,张雪,等. 改进欠抽样方法及其在非平衡数据集分类中的应用[J]. 计算机工程, 2019,45(6):218-224
[17] Ofek N., Rokach L., Stern R., et al. Fast-CBUS:A Fast Clustering-Based Undersampling Method for Addressing the Class Imbalance Problem[J]. Neurocomputing, 2017,243(Jun 21):88-102
[18] Hu J., Li Y., Yan W. X., et al. KNN-based Dynamic Query-driven Sample Rescaling Strategy for Class Imbalance Learning[J]. Neurocomputing, 2016,191(May 26):363-373
[19] 林宇,黄迅,徐凯. 基于RU-SMOTE-SVM的金融市场极端风险预警研究[J]. 预测, 2013,32(4):15-20
[20] 石洪波,陈雨文,陈鑫. SMOTE过采样及其改进算法研究综述[J]. 智能系统学报, 2019,14(6):1073-1083
[21] 刘洋. 基于不平衡数据集的机器学习算法研究[J]. 统计与决策, 2019,35(12):19-21
[22] 闫慈,田翔华,阿拉依·阿汗,等. 基于重采样技术在医学不平衡数据分类中的应用研究[J]. 中国卫生统计, 2018,35(2):177-180
[23] 魏瑾瑞,吕晓云. Logistic模型对非平衡数据的敏感性:测度、修正与比较[J]. 统计研究, 2016,33(2):79-85
[24] 罗艳虹,李治,余红梅,等. 基于代价敏感性和概率校准的先天性心脏病概率预测模型研究[J]. 中国卫生统计, 2019,36(1):36-39
[25] Thai-Nghe N., Gantner Z. Cost-Sensitive Learning Methods for Imbalanced Data[C]. International Joint Conference on Neural Networks, 2010
[26] Jin X., Xu Z., Yu Z., et al. Cost-sensitive Semi-supervised Selective Ensemble Model for Customer Credit Scoring[J]. Chinese Journal of Management Science, 2020,189(C):105118
[27] Wong M. L., Seng K., Wong P. K. Cost-sensitive Ensemble of Stacked Denoising Autoencoders for Class Imbalance Problems in Business Domain[J]. Expert Systems with Applications, 2020,141(Mar.):112918.1-112918.18
[28] Gan D., Shen J., An B., et al. Integrating TANBN with Cost Sensitive Classification Algorithm for Imbalanced Data in Medical Diagnosis[J]. Computers & Industrial Engineering, 2020,140(Feb.):106266.1-106266.9
[29] 郭海湘,黄媛玥,顾明赟,等. 基于自适应多分类器系统的甲状腺疾病诊断方法研究[J]. 系统工程理论与实践, 2018,38(8):2123-2134
[30] 马景义,吴喜之,谢邦昌. 拟自适应分类随机森林算法[J]. 数理统计与管理, 2010,29(5):805-811
[31] Timothy M. Shearman J. Varner M., et al. Modelling Post-fire Tree Mortality:Can Random Forest Improve Discrimination of Imbalanced Data?[J]. Ecological Modelling, 2019,414(15 December):108855
[32] Ma Y., Hou Y., Liu Y., et al. Research of Food Safety Risk Assessment Methods Based on Big Data[C]. IEEE International Conference on Big Data Analysis, 2016
[33] 张强,何乐平. 基于支持向量机的治安高危人员风险预警方法[J]. 管理评论, 2016,28(11):245-251
[34] 于焕杰,杜子芳. 基于随机森林的企业监管方法研究[J]. 管理世界, 2017,(9):180-181
[35] Cao P., Zhao D., Osmar Z. An Optimized Cost-sensitive SVM for Imbalanced Data Learning[C]. Pacific-asia Conference on Knowledge Discovery & Data Mining, Springer Berlin Heidelberg, 2013
[36] Yan Q., Xia S., Meng F. Optimizing Cost-Sensitive SVM for Imbalanced Data:Connecting Cluster to Classification[EB/OL]. http://doi.org/10.48550/arXiv.1702.01504,2017-02-06
[37] Jiawei H., Micheline K., Jian P. 著, 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, 2012
[38] 邓乃扬,田英杰. 支持向量机:理论、算法与拓展[M]. 北京:科学出版社, 2009
[39] 邓乃扬,田英杰. 数据挖掘中的新方法[M]. 北京:科学出版社, 2004
[40] Deng N., Tian Y., Zhang C. Support Vector Machines. Optimization Based Theory, Algorithms, and Extensions[M]. Boca Raton, Florida:CRC Press, 2012
[41] Breiman, Leo. Random Forests[J]. Machine Learning, 2001,45(1):5-32
[42] Cutler A., Cutler D. R., Stevens J. R. Random Forests[M]//Cutler A., Cutler D. R., Stevens J. R.(Eds.). Ensemble Machine Learning. Boston, MA:Springer, 2012
[43] 李建更,高志坤. 随机森林针对小样本数据类权重设置[J]. 计算机工程与应用, 2009,45(26):131-134