CSE5243 INTRO.

TO DATA MINING

Chapter 1. Introduction
Huan Sun, CSE@The Ohio State University

Slides adapted from UIUC CS412, Fall 2017, by Prof. Jiawei Han
 CSE 5243. Course Page & Schedule
 Class Homepage:
http://web.cse.ohio-state.edu/~sun.397/courses/au2017/cse5243-new.html

 Class Schedule:
9:35-10:55 AM, Tue/Thur, Bolz Hall 318

 Office hours:
 Instructor: Huan Sun @DL699, Tue 4:30-5:30PM
First week: Thur 4:30-5:30PM

 Grader: Yu Wang, wang.5205@osu.edu , By appointment

2
 CSE 5243. Textbook
 Textbook
 Jiawei Han, Micheline Kamber and Jian Pei, Data Mining: Concepts and
Techniques (3rd ed), 2011
 More resources:
https://wiki.illinois.edu//wiki/display/cs412/2.+Course+Syllabus+and+Sched
ule
 Pang-Ning Tan, Michael Steinbach, and Vipin Kumar, Introduction to
Data Mining, 2006

 Mohammed J. Zaki and Wagner Meira, Jr., Data Mining Analysis and
Concepts, 2014

 Jure Leskovec, Anand Rajaraman, Jeff Ullman, Mining of Massive

Datasets
 More resources: http://www.mmds.org/
3
 CSE 5243. Course Work and Grading
 Homework, Course Projects, and Exams
 Participation: 10% (Class attendance is critical)

 Homework: 50% (No Late Submissions!)

 Midterm exam: 20%

 Final project: 20%

 Need help and/or discussions?

 Sign on: Piazza (piazza.com/osu/autumn2017/au175243)
 Receive credits: answering questions related to the homework on Piazza and engaging
in class discussion.

4
 Chapter 1. Introduction
 What Is Data Mining?
 Why Data Mining?
 A Multi-Dimensional View of Data Mining
 What Kinds of Data Can Be Mined?
 What Kinds of Patterns Can Be Mined?
 What Kinds of Technologies Are Used?
 What Kinds of Applications Are Targeted?
 Major Issues in Data Mining
 A Brief History of Data Mining and Data Mining Society
 Summary
5
 What is Data Mining?
 Data mining (knowledge discovery from data, KDD)
 Extraction of interesting (non-trivial, implicit, previously
unknown and potentially useful) patterns or knowledge
from huge amount of data
 Data mining: a misnomer?
 Alternative names
 Knowledge discovery (mining) in databases (KDD),
knowledge extraction, data/pattern analysis, data
archeology, data dredging, information harvesting,
business intelligence, etc.

6
 Knowledge Discovery (KDD) Process
 This is a view from typical
database systems and data Pattern
warehousing communities Evaluation

 Data mining plays an essential Data

role in the knowledge discovery Mining
process
Task-relevant Data

Data Warehouse Selection

Data Cleaning
Data Integration

Databases
7
 Example: A Web Mining Framework
 Web mining usually involves
 Data cleaning
 Data integration from multiple sources
 Warehousing the data
 Data cube construction
 Data selection for data mining
 Data mining
 Presentation of the mining results
 Patterns
and knowledge to be used or stored into
knowledge-base
8
 Data Mining in Business Intelligence
Increasing potential
to support
business decisions End User
Decision
Making

Data Presentation Business

Analyst
Visualization Techniques

Data Mining Data

Information Discovery Analyst

Data Exploration
Statistical Summary, Querying, and Reporting
Data Preprocessing/Integration,
Data Warehouses
DBA
Data Sources
Paper, documents, Scientific experiments, Database Systems

9
 KDD Process: A View from ML and Statistics

Data
Input Data Data Post-
Pre- Processing
Processing Mining

Data integration Pattern discovery Pattern evaluation

Normalization Classification Pattern selection
Feature selection Clustering
Pattern interpretation
Outlier analysis
Dimension Pattern visualization
reduction …………

 This is a view from typical machine learning and statistics

communities
10
 Chapter 1. Introduction
 What Is Data Mining?
 Why Data Mining?
 A Multi-Dimensional View of Data Mining
 What Kinds of Data Can Be Mined?
 What Kinds of Patterns Can Be Mined?
 What Kinds of Technologies Are Used?
 What Kinds of Applications Are Targeted?
 Major Issues in Data Mining
 A Brief History of Data Mining and Data Mining Society
 Summary
11
 Why Data Mining?
 The Explosive Growth of Data: from terabytes to petabytes
 Data collection and data availability
 Automated data collection tools, database systems, Web,
computerized society

12
 Why Data Mining?
 The Explosive Growth of Data: from terabytes to petabytes
 Data collection and data availability
 Automated data collection tools, database systems, Web,
computerized society
 Major sources of abundant data
 Business: Web, e-commerce, transactions, stocks, …
 Science: Remote sensing, bioinformatics, scientific simulation, …
 Society and everyone: news, digital cameras, YouTube

13
 Why Data Mining?
 The Explosive Growth of Data: from terabytes to petabytes
 Data collection and data availability
 Automated data collection tools, database systems, Web,
computerized society
 Major sources of abundant data
 Business: Web, e-commerce, transactions, stocks, …
 Science: Remote sensing, bioinformatics, scientific simulation, …
 Society and everyone: news, digital cameras, YouTube
 We are drowning in data, but starving for knowledge!
 “Necessity is the mother of invention”—Data mining—Automated
analysis of massive data sets
14
 Chapter 1. Introduction
 Why Data Mining?
 What Is Data Mining?
 A Multi-Dimensional View of Data Mining
 What Kinds of Data Can Be Mined?
 What Kinds of Patterns Can Be Mined?
 What Kinds of Technologies Are Used?
 What Kinds of Applications Are Targeted?
 Major Issues in Data Mining
 A Brief History of Data Mining and Data Mining Society
 Summary
15
 Multi-Dimensional View of Data Mining
 Data to be mined
 Database data (extended-relational, object-oriented, heterogeneous), data
warehouse, transactional data, stream, spatiotemporal, time-series, sequence, text
and web, multi-media, graphs & social and information networks

16
 Multi-Dimensional View of Data Mining
 Data to be mined
 Database data (extended-relational, object-oriented, heterogeneous), data
warehouse, transactional data, stream, spatiotemporal, time-series, sequence, text
and web, multi-media, graphs & social and information networks
 Knowledge to be mined (or: Data mining functions)
 Characterization, discrimination, association, classification, clustering,
trend/deviation, outlier analysis, …
 Descriptive vs. predictive data mining
 Multiple/integrated functions and mining at multiple levels

17
 Multi-Dimensional View of Data Mining
 Data to be mined
 Database data (extended-relational, object-oriented, heterogeneous), data
warehouse, transactional data, stream, spatiotemporal, time-series, sequence, text
and web, multi-media, graphs & social and information networks
 Knowledge to be mined (or: Data mining functions)
 Characterization, discrimination, association, classification, clustering,
trend/deviation, outlier analysis, …
 Descriptive vs. predictive data mining
 Multiple/integrated functions and mining at multiple levels
 Techniques utilized
 Data-intensive, data warehouse (OLAP), machine learning, statistics, pattern
recognition, visualization, high-performance, etc.

18
 Multi-Dimensional View of Data Mining
 Data to be mined
 Database data (extended-relational, object-oriented, heterogeneous), data
warehouse, transactional data, stream, spatiotemporal, time-series, sequence, text
and web, multi-media, graphs & social and information networks
 Knowledge to be mined (or: Data mining functions)
 Characterization, discrimination, association, classification, clustering,
trend/deviation, outlier analysis, …
 Descriptive vs. predictive data mining
 Multiple/integrated functions and mining at multiple levels
 Techniques utilized
 Data-intensive, data warehouse (OLAP), machine learning, statistics, pattern
recognition, visualization, high-performance, etc.
 Applications adapted
 Retail, telecommunication, banking, fraud analysis, bio-data mining, stock market
analysis, text mining, Web mining, etc.

19
 Chapter 1. Introduction
 Why Data Mining?
 What Is Data Mining?
 A Multi-Dimensional View of Data Mining
 What Kinds of Data Can Be Mined?
 What Kinds of Patterns Can Be Mined?
 What Kinds of Technologies Are Used?
 What Kinds of Applications Are Targeted?
 Major Issues in Data Mining
 A Brief History of Data Mining and Data Mining Society
 Summary
20
 Data Mining: On What Kinds of Data?
 Database-oriented data sets and applications
 Relational database, data warehouse, transactional database
 Object-relational databases, Heterogeneous databases and legacy
databases

 Advanced data sets and advanced applications

 Data streams and sensor data
 Time-series data, temporal data, sequence data (incl. bio-sequences)
 Structure data, graphs, social networks and information networks
 Spatial data and spatiotemporal data
 Multimedia database
 Text databases

21
 The World-Wide Web
 Chapter 1. Introduction
 Why Data Mining?
 What Is Data Mining?
 A Multi-Dimensional View of Data Mining
 What Kinds of Data Can Be Mined?
 What Kinds of Patterns Can Be Mined?
 What Kinds of Technologies Are Used?
 What Kinds of Applications Are Targeted?
 Major Issues in Data Mining
 A Brief History of Data Mining and Data Mining Society
 Summary
22
 Data Mining Functions: Pattern Discovery
 Frequent patterns (or frequent itemsets)
 What items are frequently purchased together in your Walmart?

23
 Data Mining Functions: Pattern Discovery
 Frequent patterns (or frequent itemsets)
 What items are frequently purchased together in your Walmart?

 Association and Correlation Analysis

24
 Data Mining Functions: Pattern Discovery
 Frequent patterns (or frequent itemsets)
 What items are frequently purchased together in your Walmart?

 Association and Correlation Analysis

 A typical association rule

 Diaper  Beer [0.5%, 75%] (support, confidence)
 Are strongly associated items also strongly correlated?
 How to mine such patterns and rules efficiently in large datasets?
 How to use such patterns for classification, clustering, and other
applications?
25
 Data Mining Functions: Classification
 Classification and label prediction
 Construct models (functions) based on some training examples

 Describe and distinguish classes or concepts for future prediction

 Ex. 1. Classify countries based on (climate)

 Ex. 2. Classify cars based on (gas mileage)

 Predict some unknown class labels

26
 Data Mining Functions: Classification
 Classification and label prediction
 Construct models (functions) based on some training examples

 Describe and distinguish classes or concepts for future prediction

 Ex. 1. Classify countries based on (climate)

 Ex. 2. Classify cars based on (gas mileage)

 Predict some unknown class labels

 Typical methods
 Decision trees, naïve Bayesian classification, support vector
machines, neural networks, rule-based classification, pattern-
based classification, logistic regression, …

27
 Data Mining Functions: Classification
 Classification and label prediction
 Construct models (functions) based on some training examples

 Describe and distinguish classes or concepts for future prediction

 Ex. 1. Classify countries based on (climate)

 Ex. 2. Classify cars based on (gas mileage)

 Predict some unknown class labels

 Typical methods
 Decision trees, naïve Bayesian classification, support vector
machines, neural networks, rule-based classification, pattern-
based classification, logistic regression, …
 Typical applications:
 Credit card fraud detection, direct marketing, classifying stars,
diseases, web-pages, …

28
 Data Mining Functions: Cluster Analysis
 Unsupervised learning (i.e., Class
label is unknown)
 Group data to form new
categories (i.e., clusters), e.g.,
cluster houses to find distribution
patterns

29
 Data Mining Functions: Cluster Analysis
 Unsupervised learning (i.e., Class
label is unknown)
 Group data to form new
categories (i.e., clusters), e.g.,
cluster houses to find distribution
patterns
 Principle: Maximizing intra-class
similarity & minimizing interclass
similarity
 Many methods and applications

30
 Data Mining Functions: Outlier Analysis
 Outlier analysis
 Outlier: A data object that does not comply with the
general behavior of the data
 Noise or exception?―One person’s garbage could
be another person’s treasure

31
 Data Mining Functions: Outlier Analysis
 Outlier analysis
 Outlier: A data object that does not comply with the
general behavior of the data
 Noise or exception?―One person’s garbage could
be another person’s treasure
 Methods: by product of clustering or regression
analysis, …
 Useful in fraud detection, rare events analysis

32
 Data Mining Functions: Time and Ordering:
Sequential Pattern, Trend and Evolution Analysis
 Sequence, trend and evolution analysis
 Trend, time-series, and deviation analysis

 e.g., regression and value prediction

 Sequential pattern mining

 e.g., buy digital camera, then buy large

memory cards
 Periodicity analysis

 Motifs and biological sequence analysis

 Approximate and consecutive motifs

 Similarity-based analysis

 Mining data streams

 Ordered, time-varying, potentially infinite, data
streams
33
 Data Mining Functions: Structure and
Network Analysis
 Graph mining
 Finding frequent subgraphs (e.g., chemical compounds), trees (XML),
substructures (web fragments)

34
 Data Mining Functions: Structure and
Network Analysis
 Graph mining
 Finding frequent subgraphs (e.g., chemical compounds), trees (XML),
substructures (web fragments)
 Information network analysis
 Social networks: actors (objects, nodes) and relationships (edges)

 e.g., author networks in CS, terrorist networks

 Multiple heterogeneous networks

 A person could be multiple information networks: friends, family,

classmates, …
 Links carry a lot of semantic information: Link mining

35
 Data Mining Functions: Structure and
Network Analysis
 Graph mining
 Finding frequent subgraphs (e.g., chemical compounds), trees (XML),
substructures (web fragments)
 Information network analysis
 Social networks: actors (objects, nodes) and relationships (edges)

 e.g., author networks in CS, terrorist networks

 Multiple heterogeneous networks

 A person could be multiple information networks: friends, family,

classmates, …
 Links carry a lot of semantic information: Link mining

 Web mining
 Web is a big information network: from PageRank to Google

 Analysis of Web information networks

 Web community discovery, opinion mining, usage mining, …

36
 Evaluation of Knowledge
 Are all mined knowledge interesting?
 One can mine tremendous amount of “patterns”
 Some may fit only certain dimension space (time, location,
…)
 Some may not be representative, may be transient, …

37
 Evaluation of Knowledge
 Are all mined knowledge interesting?
 One can mine tremendous amount of “patterns”
 Some may fit only certain dimension space (time, location, …)
 Some may not be representative, may be transient, …
 Evaluation of mined knowledge → directly mine only interesting
knowledge?
 Descriptive vs. predictive
 Coverage
 Typicality vs. novelty
 Accuracy
 Timeliness
 …
38
 Chapter 1. Introduction
 Why Data Mining?
 What Is Data Mining?
 A Multi-Dimensional View of Data Mining
 What Kinds of Data Can Be Mined?
 What Kinds of Patterns Can Be Mined?
 What Kinds of Technologies Are Used?
 What Kinds of Applications Are Targeted?
 Major Issues in Data Mining
 A Brief History of Data Mining and Data Mining Society
 Summary
39
 Data Mining: Confluence of Multiple Disciplines

Pattern
Machine Statistics
Recognition
Learning

Applications Data Mining Visualization

Algorithm High-Performance
Database Computing
Technology

40
 Why Confluence of Multiple Disciplines?
 Tremendous amount of data
 Algorithms must be scalable to handle big data

 High-dimensionality of data
 Micro-array may have tens of thousands of dimensions

 High complexity of data

 Data streams and sensor data

 Time-series data, temporal data, sequence data

 Structure data, graphs, social and information networks

 Spatial, spatiotemporal, multimedia, text and Web data

 Software programs, scientific simulations

 New and sophisticated applications

41
 Chapter 1. Introduction
 Why Data Mining?
 What Is Data Mining?
 A Multi-Dimensional View of Data Mining
 What Kinds of Data Can Be Mined?
 What Kinds of Patterns Can Be Mined?
 What Kinds of Technologies Are Used?
 What Kinds of Applications Are Targeted?
 Major Issues in Data Mining
 A Brief History of Data Mining and Data Mining Society
 Summary
42
 Applications of Data Mining
 Web page analysis: classification, clustering, ranking
 Collaborative analysis & recommender systems
 Biological and medical data analysis
 Data mining and software engineering
 Data mining and text analysis
 Data mining and social and information network analysis
 Built-in (invisible data mining) functions in Google, MS, Yahoo!,
Linked, Facebook, …

43
 Chapter 1. Introduction
 Why Data Mining?
 What Is Data Mining?
 A Multi-Dimensional View of Data Mining
 What Kinds of Data Can Be Mined?
 What Kinds of Patterns Can Be Mined?
 What Kinds of Technologies Are Used?
 What Kinds of Applications Are Targeted?
 Major Issues in Data Mining
 A Brief History of Data Mining and Data Mining Society
 Summary
44
 Major Issues in Data Mining (1)
 Mining Methodology
 Mining various and new kinds of knowledge
 Mining knowledge in multi-dimensional space
 Data mining: An interdisciplinary effort
 Boosting the power of discovery in a networked environment
 Handling noise, uncertainty, and incompleteness of data
 Pattern evaluation and pattern- or constraint-guided mining

45
 Major Issues in Data Mining (1)
 Mining Methodology
 Mining various and new kinds of knowledge
 Mining knowledge in multi-dimensional space
 Data mining: An interdisciplinary effort
 Boosting the power of discovery in a networked environment
 Handling noise, uncertainty, and incompleteness of data
 Pattern evaluation and pattern- or constraint-guided mining
 User Interaction
 Interactive mining
 Incorporation of background knowledge
 Presentation and visualization of data mining results
46
 Major Issues in Data Mining (2)
 Efficiency and Scalability
 Efficiency and scalability of data mining algorithms
 Parallel, distributed, stream, and incremental mining methods
 Diversity of data types
 Handling complex types of data
 Mining dynamic, networked, and global data repositories
 Data mining and society
 Social impacts of data mining
 Privacy-preserving data mining

47
 Chapter 1. Introduction
 Why Data Mining?
 What Is Data Mining?
 A Multi-Dimensional View of Data Mining
 What Kinds of Data Can Be Mined?
 What Kinds of Patterns Can Be Mined?
 What Kinds of Technologies Are Used?
 What Kinds of Applications Are Targeted?
 Major Issues in Data Mining
 A Brief History of Data Mining and Data Mining Society
 Summary
48
 A Brief History of Data Mining Society
 1989 IJCAI Workshop on Knowledge Discovery in Databases
 Knowledge Discovery in Databases (G. Piatetsky-Shapiro and W.
Frawley, 1991)
 1991-1994 Workshops on Knowledge Discovery in Databases
 Advances in Knowledge Discovery and Data Mining (U. Fayyad, G.
Piatetsky-Shapiro, P. Smyth, and R. Uthurusamy, 1996)
 1995-1998 International Conferences on Knowledge Discovery in
Databases and Data Mining (KDD’95-98)
 Journal of Data Mining and Knowledge Discovery (1997)

 ACM SIGKDD conferences since 1998 and SIGKDD Explorations

 More conferences on data mining
 PAKDD (1997), PKDD (1997), SIAM-Data Mining (2001), (IEEE) ICDM
(2001), WSDM (2008), etc.
 ACM Transactions on KDD (2007)
49
 Conferences and Journals on Data Mining
 KDD Conferences  Other related conferences
 ACM SIGKDD Int. Conf. on  DB conferences: ACM SIGMOD,

Knowledge Discovery in VLDB, ICDE, EDBT, ICDT, …

Databases and Data Mining (KDD)  Web and IR conferences:

 SIAM Data Mining Conf. (SDM) WWW, SIGIR, WSDM

 (IEEE) Int. Conf. on Data Mining  ML conferences: ICML, NIPS

(ICDM)  PR conferences: CVPR,

 European Conf. on Machine  Journals

Learning and Principles and  Data Mining and Knowledge
practices of Knowledge Discovery Discovery (DAMI or DMKD)
and Data Mining (ECML-PKDD)  IEEE Trans. On Knowledge and

 Pacific-Asia Conf. on Knowledge Data Eng. (TKDE)

Discovery and Data Mining  KDD Explorations
(PAKDD)  ACM Trans. on KDD
 Int. Conf. on Web Search and

50 Data Mining (WSDM)

 Where to Find References? DBLP, CiteSeer, Google

 Data mining and KDD (SIGKDD)

 Conferences: ACM-SIGKDD, IEEE-ICDM, SIAM-DM, PKDD, PAKDD, etc.
 Journal: Data Mining and Knowledge Discovery, KDD Explorations, ACM
TKDD

 Database systems (SIGMOD)

 Conferences: ACM-SIGMOD, ACM-PODS, VLDB, IEEE-ICDE, EDBT, ICDT,
DASFAA
 Journals: IEEE-TKDE, ACM-TODS/TOIS, JIIS, J. ACM, VLDB J., Info. Sys., etc.

 AI & Machine Learning

 Conferences: Machine learning (ML), AAAI, IJCAI, COLT (Learning Theory),
CVPR, NIPS, etc.
 Journals: Machine Learning, Artificial Intelligence, Knowledge and
Information Systems, IEEE-PAMI, etc.
51
 Where to Find References? DBLP, CiteSeer, Google

 Web and IR
 Conferences: SIGIR, WWW, CIKM, etc.
 Journals: WWW: Internet and Web Information Systems

 Statistics
 Conferences: Joint Stat. Meeting, etc.
 Journals: Annals of statistics, etc.

 Visualization
 Conference proceedings: CHI, ACM-SIGGraph, etc.
 Journals: IEEE Trans. visualization and computer graphics, etc.

52
 Chapter 1. Introduction
 Why Data Mining?
 What Is Data Mining?
 A Multi-Dimensional View of Data Mining
 What Kinds of Data Can Be Mined?
 What Kinds of Patterns Can Be Mined?
 What Kinds of Technologies Are Used?
 What Kinds of Applications Are Targeted?
 Major Issues in Data Mining
 A Brief History of Data Mining and Data Mining Society
 Summary
53
 Summary
 Data mining: Discovering interesting patterns and knowledge from
massive amount of data
 A natural evolution of science and information technology, in great
demand, with wide applications
 A KDD process includes data cleaning, data integration, data
selection, transformation, data mining, pattern evaluation, and
knowledge presentation
 Mining can be performed in a variety of data
 Data mining functionalities: characterization, discrimination,
association, classification, clustering, trend and outlier analysis, etc.
 Data mining technologies and applications
 Major issues in data mining
54
 Recommended Reference Books
 Charu C. Aggarwal, Data Mining: The Textbook, Springer, 2015
 E. Alpaydin. Introduction to Machine Learning, 2nd ed., MIT Press, 2011
 R. O. Duda, P. E. Hart, and D. G. Stork, Pattern Classification, 2ed., Wiley-
Interscience, 2000
 U. Fayyad, G. Grinstein, and A. Wierse, Information Visualization in Data Mining
and Knowledge Discovery, Morgan Kaufmann, 2001
 J. Han, M. Kamber, and J. Pei, Data Mining: Concepts and Techniques. Morgan
Kaufmann, 3rd ed. , 2011
 T. Hastie, R. Tibshirani, and J. Friedman, The Elements of Statistical Learning:
Data Mining, Inference, and Prediction, 2nd ed., Springer, 2009
 T. M. Mitchell, Machine Learning, McGraw Hill, 1997
 P.-N. Tan, M. Steinbach and V. Kumar, Introduction to Data Mining, Wiley, 2005
(2nd ed. 2016)
 I. H. Witten and E. Frank, Data Mining: Practical Machine Learning Tools and
Techniques with Java Implementations, Morgan Kaufmann, 2nd ed. 2005
 Mohammed J. Zaki and Wagner Meira Jr., Data Mining and Analysis:
Fundamental Concepts and Algorithms 2014
55
56 Preview for Next Class
http://web.cse.ohio-
state.edu/~sun.397/courses/au2017/cse5243-
new.html