一种基于最大信息系数预处理的k-modes聚类方法

doi:10.16182/j.issn1004731x.joss.21-0484

系统仿真学报 ›› 2022, Vol. 34 ›› Issue (10): 2204-2212.doi: 10.16182/j.issn1004731x.joss.21-0484

一种基于最大信息系数预处理的k-modes聚类方法

李明媚¹(), 文成林¹^,²(), 胡绍林²

^1.杭州电子科技大学，浙江杭州 310018
^2.广东石油化工学院，广东茂名 525000

收稿日期:2021-05-26 修回日期:2021-08-06 出版日期:2022-10-30 发布日期:2022-10-18
通讯作者: 文成林 E-mail:851628184@qq.com;wencl@hdu.edu.cn
作者简介:李明媚(1997-)，女，硕士生，研究方向为数据挖掘。E-mail：851628184@qq.com
基金资助:
国家自然科学基金(61933013)

A K-modes Clustering Method Based on Maximal Information Coefficient Data Preprocessing

Mingmei Li¹(), Chenglin Wen¹^,²(), Shaolin Hu²

^1.Hangzhou Dianzi University, Hangzhou 310018, China
^2.Guangdong Institute of Petrochemical Technology, Maoming 525000, China

Received:2021-05-26 Revised:2021-08-06 Online:2022-10-30 Published:2022-10-18
Contact: Chenglin Wen E-mail:851628184@qq.com;wencl@hdu.edu.cn

摘要/Abstract

摘要：

为解决现有k-modes聚类方法因忽略了变量属性之间的弱相关性，常造成其在实际应用中聚类性能不佳的问题，提出一种包含属性弱相关性的新k-modes聚类方法。引入最大信息系数（maximum information coefficient， MIC)度量数据集中变量属性之间的相关性；将得到的MIC值与原有距离进行融合，建立包含属性弱相关性信息的新度量方法，以增强变量属性间相关信息的完备性，建立更加精细的k-modes聚类方法；调用3种不同的数据集，将新方法与原有的k-modes聚类方法和其他改进k-modes聚类方法的性能进行对比，并通过仿真结果表明了新方法的有效性。

关键词: 聚类方法, k-modes, 最大信息系数, 距离度量, 变量属性

Abstract:

The existing k-modes clustering method ignores the weak correlation of variable attributes, which often results in poor clustering performance in practical applications. A new k-modes clustering method that includes the weak correlation of attributes is proposed. Maximum information coefficient (MIC) is introduced to measure the correlation of variable attributes in the data set. The obtained MIC value is merged with the original distance to establish a new measurement method containing weak attribute correlation information to enhance the completeness of related information of variable attributes, and a more refined k-modes clustering method is established. Three different data sets are used to compare the performance of the new method with the existing k-modes clustering and other improved k-modes clustering methods, the simulation results shows the effectness of the new method.

Key words: clustering algorithm, k-modes, maximum information coefficient(MIC), distance metric, variable attribute

中图分类号:

TP311

李明媚, 文成林, 胡绍林. 一种基于最大信息系数预处理的k-modes聚类方法[J]. 系统仿真学报, 2022, 34(10): 2204-2212.

Mingmei Li, Chenglin Wen, Shaolin Hu. A K-modes Clustering Method Based on Maximal Information Coefficient Data Preprocessing[J]. Journal of System Simulation, 2022, 34(10): 2204-2212.

图/表 11

表1

图1

图2

图3

表2

表3

表4

表5

表6

表7

表8

参考文献 13

1	Huang Z, Ng M K.A Fuzzy K-modes Algorithm for Clustering Cate-Gorical Data[J].IEEE Transactions on Fuzzy Systems, 1999, 7(4): 446-452.
2	Ahmad A, Dey L.A Method to Compute Distance between Two Cate-Gorical Values of Same Attributein Unsupervised Learning for Categori-Cal Data Set[J].Pattern Recognition Letters(S0167-8655), 2007, 28(1): 110-118.
3	张小宇, 梁吉业, 曹付元, 等. 基于加权连接度的改进K-Modes聚类算法[J]. 广西师范大学学报(自然科学版), 2008, 26(3): 189-193.
	Zhang Xiaoyu, Liang Jiye, Cao Fuyuan, et al. An Improved K-modes Clustering Algorithm Based on Weighted Connectivity [J]. Journal of Guangxi Normal University (Natural Science Edition), 2008, 26(3): 189-193.
4	吴润秀. 基于互信息量的改进K-Modes聚类方法[J]. 统计与决策, 2012, 354(6): 89-91.
	Wu Runxiu. An Improved K-Modes Clustering Method Based on Mutual Information [J]. Statistics and Decision, 2012, 354(6): 89-91.
5	Choi Jin-Kyu, Shin Youngcheol, Hong Soonil, et al. A Study on Real-time Data Preprocessing Technique for Small Millimeter Wave Radar[J]. The Journal of The Institute of Internet, Broadcasting and Communication(S2289-0238), 2019, 19(6): 79-85.
6	Axelsson Owe, Neytcheva Maya, Ström Anders. An Efficient Preconditioning Method for State Box-Constrained Optimal Control Problems[J]. Journal of Numerical Mathematics(S1570-2820), 2018, 26(4): 185-207.
7	谢娟英, 吴肇中, 郑清泉. 基于信息增益与皮尔森相关系数的2D自适应特征选择算法[J]. 陕西师范大学学报(自然科学版), 2020, 48(6): 69-81.
	Xie Juanying, Wu Zhaozhong, Zheng Qingquan.A 2D Adaptive Feature Selection Algorithm Based on Information Gain and Pearson Correlation Coefficient [J].Journal of Shaanxi Normal University (Natural Science Edition), 2020, 48(6): 69-81.
8	彭竹清, 王艳, 纪志成. 基于改进熵值法的离散制造系统能效评价方法[J]. 系统仿真学报, 2017, 29(9): 2121-2127.
	Peng Zhuqing, Wang Yan, Ji Zhicheng. Energy Efficiency Evaluation Method of Discrete Manufacturing System Based on Improved Entropy Method[J]. Journal of System Simulation, 2017, 29(9): 2121-2127.
9	徐洪峰, 孙振强. 多标签学习中基于互信息的快速特征选择方法[J]. 计算机应用, 2019, 39(10): 2815-2821.
	Xu Hongfeng, Sun Zhenqiang. Fast Feature Selection Method Based on Mutual Information in Multi-label Learning [J]. Journal of Computer Applications, 2019, 39(10): 2815-2821.
10	叶继华, 兰清平, 刘长红, 等. 结合互信息量和Log-Gabor特征的嵌入式人脸识别[J]. 系统仿真学报, 2016, 28(9): 2214-2219.
	Ye Jihua, Lan Qingping, Liu Changhong, et al. Embedded Face Recognition Based on Mutual Information and Log-Gabor Feature [J]. Journal of System Simulation, 2016, 28(9): 2214-2219.
11	Reshef David N, Reshef Yakir A, Finucane Hilary K, et al. Detecting Novel Associations in Large Data Sets[J]. Science(S0036-8075), 2011, 334(6062): 1518-1524.
12	施振佺, 陈世平. 一种改进的k-modes聚类算法[J]. 运筹与管理, 2019, 28(12): 112-117.
	Shi Zhenquan, Chen Shiping.An Improved K-modes Clustering Algorithm [J]. Operations Research and Management, 2019, 28(12): 112-117.
13	Zhou Hongfang, Zhang Yihui, Liu Yibin. A Global-Relationship Dissimilarity Measure for the k-modes Clustering Algorithm[J]. Computational Intelligence and Neuroscience(S1687-5265), 2017: 3691316.

数据集	对象数n	属性数m
vote	435	16
cancer	699	10
mushroom	8 124	22

指标	k-modes	MI-k-modes	Z-k-modes	MIC-k-modes
AC	0.815 7	0.844 9	0.866 1	0.909 3
PE	0.828 9	0.842 3	0.862 7	0.899 5
RE	0.817 2	0.831 4	0.851 3	0.889 7

指标	k-modes	MI-k-modes	Z-k-modes	MIC-k-modes
AC	0.822 1	0.846 9	0.863 1	0.894 7
PE	0.818 3	0.840 0	0.861 1	0.882 3
RE	0.808 2	0.832 3	0.849 0	0.876 9

指标	k-modes	MI-k-modes	Z-k-modes	MIC-k-modes
AC	0.723 9	0.752 1	0.795 3	0.856 2
PE	0.751 2	0.768 9	0.802 1	0.843 7
RE	0.718 9	0.745 1	0.782 4	0.833 4

指标	对比k-modes	对比MI-k-modes	对比Z-k-modes
AC	11.475	7.622	4.987
PE	8.517	6.791	4.265
RE	8.872	7.012	4.510

一种基于最大信息系数预处理的k-modes聚类方法

A K-modes Clustering Method Based on Maximal Information Coefficient Data Preprocessing

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 11

参考文献 13

相关文章 15

编辑推荐

Metrics

本文评价

[1]	王银, 王飞翔, 孙前来. 多尺度特征融合车辆检测方法[J]. 系统仿真学报, 2022, 34(6): 1219-1229.
[2]	何俊, 洪孙焱, 周义方, 申时凯, 邹目权. 基于HMM的多维数据下扶贫对象状态预测[J]. 系统仿真学报, 2022, 34(5): 1118-1126.
[3]	周思锦, 陈棣成, 涂耿, 姜大志. 基于个性化和记忆机制的多模态情感计算模型[J]. 系统仿真学报, 2022, 34(4): 745-758.
[4]	刘永奎, 曾鸣, 张霖, 郭金维, 原思阳, 平垚垚. 基于微服务架构的云制造调度仿真系统设计与开发[J]. 系统仿真学报, 2022, 34(4): 700-711.
[5]	赵德恒, 赵应丁, 杨红云, 易文龙. 水稻叶片细胞三维可视化动态仿真[J]. 系统仿真学报, 2021, 33(10): 2432-2439.
[6]	黄晓冬, 苑海涛, 毕敬, 刘涛. 基于DQN的海战场舰船路径规划及仿真[J]. 系统仿真学报, 2021, 33(10): 2440-2448.
[7]	田智行, 陈夏铭, 姜大志. 一种离散状态与维度空间互映射的人工情感模型[J]. 系统仿真学报, 2021, 33(5): 1062-1069.
[8]	孙奕飞, 蔡亮亮, 周颐, 周忠. MRMark：面向三维视频融合的机器性能评测方法[J]. 系统仿真学报, 2020, 32(7): 1375-1384.
[9]	李楠, 张世云. 基于Web3D的海上补给虚拟装备教学训练平台[J]. 系统仿真学报, 2019, 31(6): 1136-1141.
[10]	叶学民, 林瑜茜, 卢丽芳, 李春曦. 波纹状基底上含表面活性剂液滴的聚并特征[J]. 系统仿真学报, 2018, 30(8): 2982-2990.
[11]	刘卫红, 郑啸, 程晨, 李乔. 基于TCP-nets的应急场景选择[J]. 系统仿真学报, 2018, 30(6): 2355-2361.
[12]	陈琰, 李康顺, 杨磊. 加入动态惩罚因子的GEP自动聚类算法[J]. 系统仿真学报, 2016, 28(4): 806-814.
[13]	莫尚丰, 范永健, 王晓亮, 余庆春, 李英龙. 多无线传感器网络中连接查询处理[J]. 系统仿真学报, 2016, 28(4): 833-841.
[14]	孟悦, 周明全, 税午阳, 武仲科. 基于单幅图像的三维自由曲面浮雕生成[J]. 系统仿真学报, 2015, 27(12): 3012-3017.
[15]	孙学功, 刘璟. 多物理场耦合仿真简化方法研究[J]. 系统仿真学报, 2015, 27(6): 1165-1174.

指标	对比k-modes	对比MI-k-modes	对比Z-k-modes
AC	8.883	5.644	3.661
PE	7.821	5.036	2.462
RE	8.500	5.358	3.274

指标	对比k-modes	对比MI-k-modes	对比Z-k-modes
AC	18.276	13.841	7.657
PE	12.314	9.728	5.186
RE	15.927	11.850	6.518