电子科技大学：《数据分析与数据挖掘 Data Analysis and Data Mining》课程教学资源（课件讲稿）Lecture 05 Clustering Analysis

Lecture 5 Clustering Analysis Dr.李晓瑜Xiaoyu Li Email:xiaoyuuestc@uestc.edu.cn http://blog.sciencenet.cn/u/uestc2014xiaoyu 2019-Spring SunData Group http://www.sundatagroup.org School of Information and Software Engineering,UESTC 1966 Copyright2019 by Xiaoyu Li

Dr.李晓瑜 Xiaoyu Li Email：xiaoyuuestc@uestc.edu.cn http://blog.sciencenet.cn/u/uestc2014xiaoyu 2019-Spring Lecture 5 Clustering Analysis SunData Group http://www.sundatagroup.org/ School of Information and Software Engineering, UESTC Copyright © 2019 by Xiaoyu Li. 1

●30 ata Gro10 Content (6H) 5.1 Introduction of clustering analysis 5.2 Similarity calculation 5.3 Overview of basic clustering techniques .5.4 Partitioning method 5.5 Hierarchical method 5.6 Clustering based on density and grid 5.7 Clustering based on models ·5.8 Outlier analysis 2 Copyright 2019 by Xiaoyu Li

Content（6H）  5.1 Introduction of clustering analysis  5.2 Similarity calculation  5.3 Overview of basic clustering techniques  5.4 Partitioning method  5.5 Hierarchical method  5.6 Clustering based on density and grid  5.7 Clustering based on models  5.8 Outlier analysis Copyright © 2019 by Xiaoyu Li. 2

Introduction of Clustering 4 DATA 3 Copyright 2019 by Xiaoyu Li

Copyright © 2019 by Xiaoyu Li. 3 Introduction of Clustering

1 What's Cluster Analysis?(1) ·“物以类聚，人以群分” Birds of the same feather flock together. 口 口 The result of a cluster analysis shown as the coloring of the squares into three clusters. DATA 4 Copyright 2019 by Xiaoyu Li

Copyright © 2019 by Xiaoyu Li. 4 1 What’s Cluster Analysis?(1)  “物以类聚，人以群分”  Birds of the same feather flock together.  The result of a cluster analysis shown as the coloring of the squares into three clusters

1 What's Cluster Analysis?(2) Clustering(cluster):Collection of data objects .Objects are similar in the same cluster(high intra-class similarity). Objects are different in different cluster(low inter-class similarity). Cluster Analysis Grouping the collection of physical or abstract objects into a process of multiple classes composed of similar objects. ATA 5 Copyright 2019 by Xiaoyu Li

Copyright © 2019 by Xiaoyu Li. 5  Clustering(cluster): Collection of data objects Objects are similar in the same cluster(high intra-class similarity). Objects are different in different cluster(low inter-class similarity).  Cluster Analysis Grouping the collection of physical or abstract objects into a process of multiple classes composed of similar objects. 1 What’s Cluster Analysis?(2)

1 What's Cluster Analysis?(3) Clustering is an unsupervised learning: No predefined class number Data mining function of clustering analysis .As an independent tool to obtain data distribution .As other algorithms'preprocessing steps (e.g. feature and classification) DATA 6 Copyright 2019 by Xiaoyu Li

Copyright © 2019 by Xiaoyu Li. 6  Clustering is an unsupervised learning: No predefined class number  Data mining function of clustering analysis As an independent tool to obtain data distribution As other algorithms’ preprocessing steps (e.g. feature and classification) 1 What’s Cluster Analysis?(3)

2 Typical Applications(1) 。Pattern recognition Spatial data analysis Cluster the similar regions and generate topic map in the GIS system. Exam spatial clustering and give its explanation in spatial data mining Image processing .Economics Especially in Market Research ATA 7 Copyright 2019 by Xiaoyu Li

Copyright © 2019 by Xiaoyu Li. 7 2 Typical Applications(1)  Pattern recognition  Spatial data analysis  Cluster the similar regions and generate topic map in the GIS system.  Exam spatial clustering and give its explanation in spatial data mining Image processing Economics（ Especially in Market Research ）

2 Typical Applications(2) Web Classify the documents on WEB Cluster the Web log data to find the same user access mode 。Marketing Help market analysts to find a different customer base from the customer base,so different customers can use different marketing strategy ●Earthquake research Cluster the observed epicenter points along the plate fault zone,and get the seismic risk zone ATA 8 Copyright 2019 by Xiaoyu Li

Copyright © 2019 by Xiaoyu Li. 8 2 Typical Applications(2)  Web  Classify the documents on WEB  Cluster the Web log data to find the same user access mode  Marketing  Help market analysts to find a different customer base from the customer base, so different customers can use different marketing strategy  Earthquake research  Cluster the observed epicenter points along the plate fault zone, and get the seismic risk zone

2 Typical Applications(3) ●Land use In the database of earth monitoring,the same land use region is found ·Insurance industry The customer base of the higher claim rate in automobile insurance is found 。Urban planning Group it according to the type of house,value and location ATA 9 Copyright 2019 by Xiaoyu Li

Copyright © 2019 by Xiaoyu Li. 9 2 Typical Applications(3)  Land use  In the database of earth monitoring, the same land use region is found  Insurance industry  The customer base of the higher claim rate in automobile insurance is found  Urban planning  Group it according to the type of house, value and location

3 What's Good Clustering Analysis? High class internal similarity; Low class similarity; As a branch of statistics,the clustering analysis researchi theme is mainly based 0n distance- clustering a high-quality clustering analysis result will be decided on the used clustering method; The implementation of the similarity measure and the method used in clustering method; Ability to discover hidden patterns. ATA 10 Copyright 2019 by Xiaoyu Li

Copyright © 2019 by Xiaoyu Li. 10 3 What’s Good Clustering Analysis?  High class internal similarity;  Low class similarity;  As a branch of statistics, the clustering analysis research theme is mainly based on distance￾clustering ; a high-quality clustering analysis result will be decided on the used clustering method;  The implementation of the similarity measure and the method used in clustering method;  Ability to discover hidden patterns