jointly with Jiawei Han a Data Mining Summer Course in Shanghai, China in.
July 1998. He has ... slides used for a Data Mining course at Univ. of Munich
jointly taught with. Prof. Dr. Martin Ester (now teaching ..... M. H. Dunham. Data
Mining: ...
Data Management and Exploration Prof. Dr. Thomas Seidl
Data Mining Algorithms Lecture Course with Tutorials Wintersemester 2003/04 RWTH Aachen, Lehrstuhl für Informatik IX – Data Management and Exploration – Prof. Dr. Thomas Seidl
Data Management and Exploration Prof. Dr. Thomas Seidl
Course Organization (1)
Weekly Schedule
Tuesday, 13:45 – 15:15 h, AH VI (Lecture)
Thursday, 13:45 – 15:15 h, AH VI (Lecture)
Friday, 10:00 – 11:30 h, AH VI (Tutorials) Begins: Oct. 24th
People
Prof. Dr. Thomas Seidl Office hour: Monday, 13:00 – 14:00 h
Dipl.-Inform. Ira Assent
Dipl.-Inform. Christoph Brochhaus
WS 2003/04
Data Mining Algorithmen
2
Data Management and Exploration Prof. Dr. Thomas Seidl
Course Organization (2)
Credits
Lecture course with tutorials (4 + 2 hrs/wk) 8 ECTS for “Praktische Informatik” or specialization “Datenbanken und Datenexploration” / “Databases and Data Mining”
Selected exercises as homework (50% points for exam admission)
Oral or written exam at the end of semester
Material (Slides, Exercises, etc.) available from: http://www-i9.informatik.rwth-aachen.de/Lehre/ DataMiningAlgorithmen/
WS 2003/04
3
Data Mining Algorithmen
Data Management and Exploration Prof. Dr. Thomas Seidl
Content of the Course
Introduction
Data warehousing and OLAP for data mining
Data preprocessing
Clustering
Classification and prediction techniques
Mining association rules in large databases
Further topics
WS 2003/04
Data Mining Algorithmen
4
Data Management and Exploration Prof. Dr. Thomas Seidl
Textbook / Acknowledgements The slides used in this course are modified versions of the copyrighted original slides provided by the authors of the adopted textbooks: © Jiawei Han and Micheline Kamber: Data Mining – Concepts and Techniques Morgan Kaufmann Publishers, 2000. http://www.cs.sfu.ca/~han/dmbook © Martin Ester and Jörg Sander: Knowledge Discovery in Databases – Techniken und Anwendungen Springer Verlag, 2000 (in German). WS 2003/04
5
Data Mining Algorithmen
Data Management and Exploration Prof. Dr. Thomas Seidl
Source and History of the Slides
Prof. Dr. Jiawei Han started working on this set of slides for an UCLA Extension course in February 1998. Dr. Hongjun Lu from Hong Kong Univ. of Science and Technology taught jointly with Jiawei Han a Data Mining Summer Course in Shanghai, China in July 1998. He has contributed many slides to it. Some graduate students have contributed many new slides in the following years. Notable contributors include Eugene Belchev, Jian Pei, and Osmar R. Zaiane (now teaching in Univ. of Alberta). Prof. Dr. Jörg Sander, Univ. of Alberta in Edmonton, Canada provided some more slides, particularly on clustering algorithms. They are an extension of slides used for a Data Mining course at Univ. of Munich jointly taught with Prof. Dr. Martin Ester (now teaching in Simon-Fraser-Univ. in Vancouver).
WS 2003/04
Data Mining Algorithmen
6
Data Management and Exploration Prof. Dr. Thomas Seidl
Chapter 1. Introduction
Motivation: Why data mining?
What is data mining?
Data Mining: On what kind of data?
Data mining functionality
Are all the patterns interesting?
Classification of data mining systems
Major issues in data mining
WS 2003/04
7
Data Mining Algorithmen
Data Management and Exploration Prof. Dr. Thomas Seidl
Motivation: “Necessity is the Mother of Invention”
Data explosion problem
Automated data collection tools and mature database technology lead to tremendous amounts of data stored in databases, data warehouses and other information repositories
We are drowning in data, but starving for knowledge!
Solution: Data warehousing and data mining
Data Warehousing and on-line analytical processing (OLAP)
Extraction of interesting knowledge (rules, regularities, patterns, constraints) from data in large databases
WS 2003/04
Data Mining Algorithmen
8
Data Management and Exploration Prof. Dr. Thomas Seidl
Evolution of Database Technology
1960s:
1970s:
Relational data model, relational DBMS implementation
1980s:
Data collection, database creation, IMS and network DBMS
RDBMS, advanced data models (extended-relational, OO, deductive, etc.) and application-oriented DBMS (spatial, scientific, engineering, etc.)
1990s—2000s:
Data mining and data warehousing, multimedia databases, and Web databases
WS 2003/04
9
Data Mining Algorithmen
Data Management and Exploration Prof. Dr. Thomas Seidl
What Is Data Mining?
Data mining (knowledge discovery in databases):
Alternative names and their “inside stories”:
Data mining: a misnomer? Knowledge discovery(mining) in databases (KDD), knowledge extraction, data/pattern analysis, data archeology, data dredging (“Ausbaggern”), information harvesting, business intelligence, etc.
What is not data mining?
WS 2003/04
Extraction of interesting (non-trivial, implicit, previously unknown and potentially useful) information or patterns from data in large databases
(Deductive) query processing. Expert systems or small ML/statistical programs Data Mining Algorithmen
10
Why Data Mining? — Potential Applications
Data Management and Exploration Prof. Dr. Thomas Seidl
Database analysis and decision support
Market analysis and management
Risk analysis and management
target marketing, customer relation management, market basket analysis, cross selling, market segmentation Forecasting, customer retention (Eigenbeteiligung), improved underwriting, quality control, competitive analysis
Fraud detection and management
Other Applications
Text mining (news group, email, documents) and Web analysis.
Intelligent query answering
WS 2003/04
11
Data Mining Algorithmen
Data Management and Exploration Prof. Dr. Thomas Seidl
Market Analysis and Management (1)
Where are the data sources for analysis?
Target marketing
Find clusters of “model” customers who share the same characteristics: interest, income level, spending habits, etc.
Determine customer purchasing patterns over time
Credit card transactions, loyalty cards, discount coupons, customer complaint calls, plus (public) lifestyle studies
Conversion of a single to a joint bank account: marriage, etc.
Cross-market analysis
WS 2003/04
Associations/co-relations between product sales
Prediction based on the association information Data Mining Algorithmen
12
Data Management and Exploration Prof. Dr. Thomas Seidl
Market Analysis and Management (2)
Customer profiling
data mining can tell you what types of customers buy what products (clustering or classification)
Identifying customer requirements
identifying the best products for different customers
use prediction to find what factors will attract new customers
Provide summary information
various multidimensional summary reports
statistical summary information (data central tendency and variation of the data)
WS 2003/04
13
Data Mining Algorithmen
Data Management and Exploration Prof. Dr. Thomas Seidl
Corporate Analysis and Risk Management
Finance planning and asset evaluation
Resource planning:
cash flow analysis and prediction contingent claim analysis to evaluate assets cross-sectional and time series analysis (financial-ratio, trend analysis, etc.) summarize and compare the resources and spending
Competition:
WS 2003/04
monitor competitors and market directions group customers into classes and a class-based pricing procedure set pricing strategy in a highly competitive market Data Mining Algorithmen
14
Data Management and Exploration Prof. Dr. Thomas Seidl
Fraud Detection and Management
Applications
Approach
widely used in health care, retail, credit card services, telecommunications (phone card fraud), etc. use historical data to build models of fraudulent behavior and use data mining to help identify similar instances
Examples
auto insurance: detect a group of people who stage accidents to collect on insurance money laundering: detect suspicious money transactions (e.g., US Treasury's Financial Crimes Enforcement Network) medical insurance: detect professional patients and ring of doctors and ring of references
WS 2003/04
15
Data Mining Algorithmen
Data Management and Exploration Prof. Dr. Thomas Seidl
Data Mining and Business Intelligence Increasing potential to support business decisions
Making Decisions Data Presentation Visualization Techniques Data Mining Information Discovery
End User
Business Analyst Data Analyst
Data Exploration Statistical Analysis, Querying and Reporting Data Warehouses / Data Marts OLAP, MDA Data Sources Paper, Files, Information Providers, Database Systems, OLTP WS 2003/04
Data Mining Algorithmen
DBA
16
Data Management and Exploration Prof. Dr. Thomas Seidl
Architecture of Typical Data Mining Systems Graphical user interface
Pattern evaluation Data mining engine Database or data warehouse server Data cleaning & data integration
Databases WS 2003/04
Knowledge-base Filtering
Data Warehouse 17
Data Mining Algorithmen
Data Management and Exploration Prof. Dr. Thomas Seidl
Data Mining: On What Kind of Data?
Relational databases Data warehouses Transactional databases Advanced DB and information repositories
WS 2003/04
Object-oriented and object-relational databases Spatial databases Time-series data and temporal data Text databases and multimedia databases Heterogeneous and legacy databases WWW Data Mining Algorithmen
18
Data Management and Exploration Prof. Dr. Thomas Seidl
The KDD Process
Selections Projections Transformations
Data Mining Task-relevant Data
Visualization Evaluation Patterns
Knowledge
“Data Warehouse” Data Cleaning Data Integration
Data Mining: • a central step in the process • application of algorithms that find “patterns” in the data.
Databases/ Information Repositories WS 2003/04
19
Data Mining Algorithmen
Data Management and Exploration Prof. Dr. Thomas Seidl
Steps of a KDD Process
Learning the application domain:
Creating a target data set: data selection Data cleaning and preprocessing: (may take 60% of effort!) Data reduction and transformation:
summarization, classification, regression, association, clustering.
Choosing the mining algorithm(s) Data mining: search for patterns of interest Pattern evaluation and knowledge presentation
Find useful features, dimensionality/variable reduction, invariant representation.
Choosing functions of data mining
relevant prior knowledge and goals of application
visualization, transformation, removing redundant patterns, etc.
Use of discovered knowledge
WS 2003/04
Data Mining Algorithmen
20
Data Management and Exploration Prof. Dr. Thomas Seidl
Data Cleaning and Integration
Integration of data from different sources
Mapping of attribute names Data Cleaning (e.g. C_Nr → O_Id) Data Integration “Data Warehouse” Joining different tables (e.g. Table1 = [C_Nr, Info1] Databases/ Information Repositories and Table2 = [O_Id, Info2] ⇒ JoinedTable = [O_Id, Info1, Info2])
Elimination of inconsistencies Elimination of noise Computation of Missing Values (if necessary and possible)
Fill in missing values by some strategy (e.g. default value, average value, or application specific computations)
WS 2003/04
21
Data Mining Algorithmen
Data Management and Exploration Prof. Dr. Thomas Seidl
Focusing on task-relevant data
Selections
Select the relevant tuples/rows from the database tables
Selections Projections Transformations
(e.g., sales data for the year 2001)
Projections
Task-relevant Data
“Data Warehouse”
Select the relevant attributes/columns from the database tables (e.g., “id”, “date” “amount” from (Id, name, date, location, amount))
Transformations, e.g.:
Normalization (e.g., age:[18, 87] Æ n_age:[0, 100]
Discretisation of numerical attributes
Computation of derived tuples/rows and derived attributes
(e.g., amount:[0, 100] Æ d_amount:{low, medium, high}
WS 2003/04
aggregation of sets of tuples, e.g., total amount per months new attributes, e.g., diff = sales current month – sales previous month) Data Mining Algorithmen
22
Data Management and Exploration Prof. Dr. Thomas Seidl
Basic Data Mining Tasks Data
Mining Clustering Task-relevant Patterns Data Classification Association Rules Concept Characterization and Discrimination Other methods
Outlier detection Sequential patterns Trends and analysis of changes Methods for special data types, e.g., spatial data mining, web mining …
WS 2003/04
23
Data Mining Algorithmen
Data Management and Exploration Prof. Dr. Thomas Seidl
Clustering
Class labels are unknown: Group objects into sub-groups (clusters)
Similarity function (or dissimilarity fct. = distance) to measure similarity between objects Objective: “maximize” intra-class similarity and “minimize” interclass similarity
Applications
Customer profiling/segmentation Document or image collections Web access patterns ...
WS 2003/04
Data Mining Algorithmen
24
Data Management and Exploration Prof. Dr. Thomas Seidl
Classification Class labels are known for a small set of “training data”: Find models/functions/rules (based on attribute values of the training examples) that
describe and distinguish classes predict class membership for “new” objects
Applications
a a
a
aa ba a
b b b ab b a b b b
Classify gene expression values for tissue samples to predict disease type and suggest best possible treatment Automatic assignment of categories to large sets of newly observed celestial objects Predict unknown or missing values (Æ KDD pre-processing step) ...
WS 2003/04
25
Data Mining Algorithmen
Data Management and Exploration Prof. Dr. Thomas Seidl
Association Rules Find frequent associations in transaction databases
Frequently co-occurring items in the set of transactions (frequent itemsets): indicate correlations or causalities Rules with a minimum confidence Transaction ID Items Bought based on frequent itemsets
Applications:
2000
A,B,C
1000
A,C
Market-basket analysis 4000 Cross-marketing 5000 Catalog design Also used as a basis for clustering, classification
A,D B,E,F
Examples: buys(x, “diapers”) → buys(x, “beers”) [support: 0.5%, confidence: 60%] major(x, “CS”) ^ takes(x, “DB”) → grade(x, “A”) [support: 1%, confidence: 75%] WS 2003/04
Data Mining Algorithmen
26
Data Management and Exploration Prof. Dr. Thomas Seidl
Generalization, Characterization, Discrimination
Generalize, summarize, and contrast data characteristics
Based on attribute aggregation along concept hierarchies Data cube approach (OLAP) Attribute-oriented induction approach all
all Europe
region
country
city
Germany
Frankfurt
office WS 2003/04
...
North_America
...
Spain
Canada
Vancouver
...
L. Chan
...
...
...
Mexico
Toronto
M. Wind 27
Data Mining Algorithmen
Data Management and Exploration Prof. Dr. Thomas Seidl
Other Methods
Outlier detection
Trends and Evolution Analysis
Sequential patterns (find re-occurring sequences of events) Regression analysis
Methods for special data types, and applications e.g.,
Find objects that do not comply with the general behavior of the data (fraud detection, rare events analysis)
Spatial data mining Web mining Bio-KDD Graphs ...
…
WS 2003/04
Data Mining Algorithmen
28
Data Management and Exploration Prof. Dr. Thomas Seidl
Evaluation and Visualization
Integration of visualization and data mining
data visualization data mining result visualization data mining process visualization interactive visual data mining
Visualization Evaluation Patterns
Knowledge
Different types of 2D/3D plots, charts and diagrams are used, e.g.:
WS 2003/04
Box-plots Trees X-Y-Plots Parallel coordinates ... 29
Data Mining Algorithmen
Data Management and Exploration Prof. Dr. Thomas Seidl
Are All the “Discovered” Patterns Interesting?
A data mining system/query may generate thousands of patterns, not all of them are interesting.
Interestingness measures:
Suggested approach: Human-centered, query-based, focused mining A pattern is interesting if it is easily understood by humans, valid on new or test data with some degree of certainty, potentially useful, novel, or validates some hypothesis that a user seeks to confirm
Objective vs. subjective interestingness measures:
WS 2003/04
Objective: based on statistics and structures of patterns, e.g., support, confidence, etc. Subjective: based on user’s belief in the data, e.g., unexpectedness, novelty, actionability, etc. Data Mining Algorithmen
30
Data Management and Exploration Prof. Dr. Thomas Seidl
Can We Find All and Only Interesting Patterns?
Find all the interesting patterns: Completeness
Can a data mining system find all the interesting patterns?
Association vs. classification vs. clustering
Search for only interesting patterns: Optimization
Can a data mining system find only the interesting patterns?
Approaches
First generate all the patterns and then filter out the uninteresting ones. Generate only the interesting patterns—mining query optimization
WS 2003/04
31
Data Mining Algorithmen
Data Management and Exploration Prof. Dr. Thomas Seidl
Data Mining: Confluence of Multiple Disciplines Database Technology
Machine Learning
Statistics
Data Mining
Information Science WS 2003/04
Visualization
Other Disciplines Data Mining Algorithmen
32
Data Management and Exploration Prof. Dr. Thomas Seidl
Data Mining: Classification Schemes
WS 2003/04
General functionality
Descriptive data mining
Predictive data mining
Different views, different classifications
Kinds of databases to be mined
Kinds of knowledge to be discovered
Kinds of techniques utilized
Kinds of applications adapted 33
Data Mining Algorithmen
Data Management and Exploration Prof. Dr. Thomas Seidl
A Multi-Dimensional View of Data Mining Classification
Databases to be mined Relational, transactional, object-oriented, object-relational, active, spatial, time-series, text, multi-media, heterogeneous, legacy, WWW, etc. Knowledge to be mined Characterization, discrimination, association, classification, clustering, trend, deviation and outlier analysis, etc. Multiple/integrated functions and mining at multiple levels Techniques utilized Database-oriented, data warehouse (OLAP), machine learning, statistics, visualization, neural network, etc. Applications adapted
WS 2003/04
Retail, telecommunication, banking, fraud analysis, DNA mining, stock market analysis, Web mining, Weblog analysis, etc. Data Mining Algorithmen
34
Data Management and Exploration Prof. Dr. Thomas Seidl
OLAP Mining: An Integration of Data Mining and Data Warehousing
Data mining systems, DBMS, Data warehouse systems coupling
On-line analytical mining data
integration of mining and OLAP technologies
Interactive mining multi-level knowledge
No coupling, loose-coupling, semi-tight-coupling, tight-coupling
Necessity of mining knowledge and patterns at different levels of abstraction by drilling/rolling, pivoting, slicing/dicing, etc.
Integration of multiple mining functions
WS 2003/04
Characterized classification, first clustering and then association 35
Data Mining Algorithmen
Data Management and Exploration Prof. Dr. Thomas Seidl
Major Issues in Data Mining (1)
Mining methodology and user interaction
Mining different kinds of knowledge in databases
Interactive mining of knowledge at multiple levels of abstraction
Incorporation of background knowledge
Data mining query languages and ad-hoc data mining
Expression and visualization of data mining results
Handling noise and incomplete data
Pattern evaluation: the interestingness problem
Performance and scalability
Efficiency and scalability of data mining algorithms
Parallel, distributed and incremental mining methods
WS 2003/04
Data Mining Algorithmen
36
Data Management and Exploration Prof. Dr. Thomas Seidl
Major Issues in Data Mining (2) Issues relating to the diversity of data types
Handling relational and complex types of data Mining information from heterogeneous databases and global information systems (WWW)
Issues related to applications and social impacts
Application of discovered knowledge Domain-specific data mining tools Intelligent query answering Process control and decision making Integration of the discovered knowledge with existing knowledge: A knowledge fusion problem Protection of data security, integrity, and privacy
WS 2003/04
37
Data Mining Algorithmen
Data Management and Exploration Prof. Dr. Thomas Seidl
Summary
Data mining: discovering interesting patterns from large amounts of data A natural evolution of database 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 information repositories Data mining functionalities: characterization, discrimination, association, classification, clustering, outlier and trend analysis, etc.
Classification of data mining systems
Major issues in data mining
WS 2003/04
Data Mining Algorithmen
38
A Brief History of Data Mining Society
1989 IJCAI (Int‘l Joint Conf. on Artificial Intelligence) Workshop on Knowledge Discovery in Databases (Piatetsky-Shapiro)
Advances in Knowledge Discovery and Data Mining (U. Fayyad, G. PiatetskyShapiro, P. Smyth, and R. Uthurusamy, 1996)
1995-1998 International Conferences on Knowledge Discovery in Databases and Data Mining (KDD’95-98)
Knowledge Discovery in Databases (G. Piatetsky-Shapiro and W. Frawley, 1991)
1991-1994 Workshops on Knowledge Discovery in Databases
Data Management and Exploration Prof. Dr. Thomas Seidl
Journal of Data Mining and Knowledge Discovery (1997)
1998 ACM SIGKDD, SIGKDD’1999-2001 conferences, and SIGKDD Explorations More conferences on data mining
PAKDD, PKDD, SIAM-Data Mining, (IEEE) ICDM, etc.
WS 2003/04
39
Data Mining Algorithmen
Data Management and Exploration Prof. Dr. Thomas Seidl
Where to Find References?
Data mining and KDD (SIGKDD member CDROM):
Database field (SIGMOD member CD ROM):
Conference proceedings: Machine learning, AAAI, IJCAI, etc. Journals: Machine Learning, Artificial Intelligence, etc.
Statistics:
Conference proceedings: ACM-SIGMOD, ACM-PODS, VLDB, ICDE, EDBT, DASFAA Journals: ACM-TODS, J. ACM, IEEE-TKDE, JIIS, etc.
AI and Machine Learning:
Conference proceedings: KDD, and others, such as PKDD, PAKDD, etc. Journal: Data Mining and Knowledge Discovery
Conference proceedings: Joint Stat. Meeting, etc. Journals: Annals of statistics, etc.
Visualization:
WS 2003/04
Conference proceedings: CHI (Comp. Human Interaction), etc. Journals: IEEE Trans. visualization and computer graphics, etc. Data Mining Algorithmen
40
Data Management and Exploration Prof. Dr. Thomas Seidl
References
U. M. Fayyad, G. Piatetsky-Shapiro, P. Smyth, and R. Uthurusamy. Advances in Knowledge Discovery and Data Mining . AAAI/MIT Press, 1996. J. Han and M. Kamber. Data Mining: Concepts and Techniques. Morgan Kaufmann, 2000. T. Imielinski and H. Mannila. A database perspective on knowledge discovery. Communications of the ACM, 39:58-64, 1996. G. Piatetsky-Shapiro, U. Fayyad, and P. Smith. From data mining to knowledge discovery: An overview. In U.M. Fayyad, et al. (eds.), Advances in Knowledge Discovery and Data Mining, 1-35. AAAI/MIT Press, 1996. G. Piatetsky-Shapiro and W. J. Frawley. Knowledge Discovery in Databases. AAAI/MIT Press, 1991. M. Ester and J. Sander. Knowledge Discovery in Databases: Techniken und Anwendungen. Springer Verlag, 2000 (in German). M. H. Dunham. Data Mining: Introductory and Advanced Topics. Prentice Hall, 2003. D. Hand, H. Mannila, P. Smyth. Principles of Data Mining. MIT Press, 2001.
WS 2003/04
Data Mining Algorithmen
41