1.
A.V.C.COLLEGE OF ENGINEERING
MANNAMPANDAL, MAYILADUTHURAI-609 305
COURSE MATERIAL
FOR THE SUBJECT OF
SUB NAME : CS1011 DATA WAREHOUSING AND DATA MINING
SEM : VII
DEPARTMENT : COMPUTER SCIENCE AND ENGINEERING
ACADEMIC YEAR : 2013-2013
NAME OF THE FACULTY : PARVATHI.M
DESIGNATION : Asst.Professor
1
2.
A.V.C College of Engineering
Department of Computer Science & Engineering
2013 Odd Semester
Lesson Plan
SYLLABUS
ELECTIVE II
CS1011 – DATA WAREHOUSING AND DATA MINING L T P
3 0 0
UNIT I BASICS OF DATA WAREHOUSING 8
Introduction − Data warehouse − Multidimensional data model − Data warehouse
architecture −Implementation − Further development − Data warehousing to data mining.
UNIT II DATA PREPROCESSING, LANGUAGE, ARCHITECTURES,
CONCEPT
DESCRIPTION 8
Why preprocessing − Cleaning − Integration − Transformation − Reduction −
Discretization – Concept hierarchy generation − Data mining primitives − Query
language − Graphical user interfaces − Architectures − Concept description − Data
generalization − Characterizations − Class comparisons − Descriptive statistical
measures.
UNIT III ASSOCIATION RULES 9
Association rule mining − Single-dimensional boolean association rules from
transactional databases − Multi level association rules from transaction databases
UNIT IV CLASSIFICATION AND CLUSTERING 12
Classification and prediction − Issues − Decision tree induction − Bayesian classification
– Association rule based − Other classification methods − Prediction − Classifier
accuracy − Cluster analysis – Types of data − Categorization of methods − Partitioning
methods − Outlier analysis.
UNIT V RECENT TRENDS 8
Multidimensional analysis and descriptive mining of complex data objects − Spatial
databases −
Multimedia databases − Time series and sequence data − Text databases − World Wide
Web −
Applications and trends in data mining.
Total:
45
2
3.
TEXT BOOKS
1. Han, J. and Kamber, M., “Data Mining: Concepts and Techniques”, Harcourt India /
Morgan
Kauffman, 2001.
2. Margaret H. Dunham, “Data Mining: Introductory and Advanced Topics”, Pearson
Education
2004.
REFERENCES
1. Sam Anahory and Dennis Murry, “Data Warehousing in the real world”, Pearson
Education,
2003.
2. David Hand, Heikki Manila and Padhraic Symth, “Principles of Data Mining”, PHI
2004.
3. W.H.Inmon, “Building the Data Warehouse”, 3rd Edition, Wiley, 2003.
4. Alex Bezon and Stephen J.Smith, “Data Warehousing, Data Mining and OLAP”,
McGraw- Hill Edition, 2001.
5. Paulraj Ponniah, “Data Warehousing Fundamentals”, Wiley-Interscience Publication,
2003.
3
4.
UNIT I BASICS OF DATA WAREHOUSING
Introduction − Data warehouse − Multidimensional data model − Data warehouse
architecture −Implementation – Fur ther development − Data warehousing to data mining.
1.1Introduction to Data Warehousing
A data warehouse is a collection of data marts representing historical data from
different operations in the company. This data is stored in a structure optimized
for querying and data analysis as a data warehouse. Table design, dimensions
and organization should be consistent throughout a data warehouse so that
reports or queries across the data warehouse are consistent. A data warehouse
can also be viewed as a database for historical data from different functions
within a company.
Bill Inmon coined the term Data Warehouse in 1990, which he defined in the
following way: "A warehouse is a subject-oriented, integrated, time-variant
and non-volatile collection of data in support of management's decision making
process".
• Subject Oriented: Data that gives information about a particular subject
instead of about a company's ongoing operations. Focusing on the
modelling and analysis of data for decision makers, not on daily
operations or transaction processing. It is used to provide a simple and
concise view around particular subject issues by excluding data that are
not useful in the decision support process.
• Integrated: Data that is gathered into the data warehouse from a variety
of sources and merged into a coherent whole. Data cleaning and data
integration techniques are applied. It is used to ensure consistency in
naming conventions, encoding structures, attribute measures, etc.
among different data sources. E.g., Hotel price: currency, tax, breakfast
covered, etc.
• Time-variant: All data in the data warehouse is identified with a
particular time period. The time horizon for the data warehouse is
significantly longer than that of operational systems.
o Operational database: current value data.
o Data warehouse data: provide information from a historical
perspective (e.g., past 5-10 years)
• Non-volatile: Data is stable in a data warehouse. More data is added
but data is never removed. Operational update of data does not occur in
the data warehouse environment. It does not require transaction
processing, recovery, and concurrency control mechanisms. It requires
only two operations in data accessing:
4
5.
o initial loading of data and
o Access of data.
Data Warehouse is a single, complete and consistent store of data obtained
from a variety of different sources made available to end users in what they can
understand and use in a business context.
It can be
• Used for decision Support
• Used to manage and control business
• Used by managers and end-users to understand the business and make
judgments
Data Warehousing is an architectural construct of information systems that
provides users with current and historical decision support information that is
hard to access or present in traditional operational data stores
Other important terminology
Enterprise Data warehouse: It collects all information about subjects
(customers, products, sales, assets, personnel) that span the entire organization
Data Mart: Departmental subsets that focus on selected subjects. A data mart is
a segment of a data warehouse that can provide data for reporting and analysis
on a section, unit, department or operation in the company. E.g. sales, payroll,
production. Data marts are sometimes complete individual data warehouses
which are usually smaller than the corporate data warehouse.
Decision Support System (DSS): Information technology to help the
knowledge worker (executive, manager, and analyst) makes faster & better
decisions
Drill-down: Traversing the summarization levels from highly summarized data
to the underlying current or old detail
Metadata: Data about data. It is used to describe location and description of
warehouse system components such as names, definition, structure…
Benefits of data warehousing
• Data warehouses are designed to perform well with aggregate queries
running on large amounts of data.
• The structure of data warehouses is easier for end users to navigate,
understand and query against unlike the relational databases primarily
designed to handle lots of transactions.
5
6.
• Data warehouses enable queries that cut across different segments of a
company's operation. E.g. production data could be compared against
inventory data even if they were originally stored in different databases
with different structures.
• Queries that would be complex in normalized databases could be easier to
build and maintain in data warehouses, decreasing the workload on
transaction systems.
• Data warehousing is an efficient way to manage and report on data that is
from a variety of sources, non uniform and scattered throughout a company.
• Data warehousing is an efficient way to manage demand for lots of
information from lots of users.
• Data warehousing provides the capability to analyze large amounts of
historical data for nuggets of wisdom that can provide an organization with
competitive advantage.
Operational and informational Data
1. Operational Data are used for focusing on transactional function such as bank
card withdrawals and deposits and they are
• Detailed
• Updateable
• Reflects current data
2. Informational Data are used for focusing on providing answers to problems
posed by decision makers
• Summarized
• Non updateable
1.1 Building a Data Warehouse
The selection of data warehouse technology - both hardware and software -
depends on many factors, such as:
• the volume of data to be accommodated,
• the speed with which data is needed,
• the history of the organization,
• which level of data is being built,
• how many users there will be,
• what kind of analysis is to be performed,
• Cost of technology, etc.
The hardware is typically mainframe, parallel, or client/server hardware. The
software that must be selected is for the basic data base manipulation of the data as
it resides on the hardware. Typically the software is either full function DBMS or
specialized data base software that has been optimized for the data warehouse.
Other software that needs to be considered is the interface software that provides
transformation and metadata capability such as PRISM Solutions Warehouse
6
7.
Manager. A final piece of software that is important is the software needed for
changed data capture.
A rough sizing of data needs to be done to determine the fitness of the hardware
and software platforms. If the hardware and DBMS software are much too large
for the data warehouse, the costs of building and running the data warehouse will
be exorbitant. Even though performance will be no problem, development and
operational costs and finances will be a problem.
Conversely, if the hardware and DBMS software are much too small for the size of
the data warehouse, then performance of operations and the ultimate end user
satisfaction with the data warehouse will suffer. So, it is important that there be a
comfortable fit between the data warehouse and the hardware and DBMS software
that will house and manipulate the warehouse.
There are two factors required to build and use data warehouse. They are:
Business factors:
• Business users want to make decision quickly and correctly using all
available data.
Technological factors:
• To address the incompatibility of operational data stores
• IT infrastructure is changing rapidly. Its capacity is increasing and cost is
decreasing so that building a data warehouse is easy
There are several things to be considered while building a successful data
warehouse
1.2.1 Business considerations:
Organizations interested in development of a data warehouse can choose one of the
following two approaches:
• Top - Down Approach (Suggested by Bill Inmon)
• Bottom - Up Approach (Suggested by Ralph Kimball)
a. Top - Down Approach
In the top down approach suggested by Bill Inmon, we build a centralized
repository to house corporate wide business data. This repository is called
Enterprise Data Warehouse (EDW). The data in the EDW is stored in a normalized
form in order to avoid redundancy.
The central repository for corporate wide data helps us maintain one version of
truth of the data. The data in the EDW is stored at the most detail level. The reason
to build the EDW on the most detail level is to leverage the flexibility to be used
by multiple departments and to cater for future requirements.
7
8.
The disadvantages of storing data at the detail level are
1. The complexity of design increases with increasing level of detail.
2. It takes large amount of space to store data at detail level, hence increased
cost.
Once the EDW is implemented we start building subject area specific data marts
which contain data in a de normalized form also called star schema. The data in the
marts are usually summarized based on the end users analytical requirements.
The reason to de normalize the data in the mart is to provide faster access to the
data for the end users analytics. If we were to have queried a normalized schema
for the same analytics, we would end up in a complex multiple level joins that
would be much slower as compared to the one on the de normalized schema.
The top-down approach can be used when
1. The business has complete clarity on all or multiple subject areas data
warehouse requirements.
2. The business is ready to invest considerable time and money.
The advantage of using the Top Down approach is that we build a centralized
repository to cater for one version of truth for business data. This is very important
for the data to be reliable, consistent across subject areas and for reconciliation in
case of data related contention between subject areas.
The disadvantage of using the Top Down approach is that it requires more time
and initial investment. The business has to wait for the EDW to be implemented
followed by building the data marts before which they can access their reports.
b. Bottom Up Approach
The bottom up approach suggested by Ralph Kimball is an incremental approach
to build a data warehouse. In this approach data marts are built separately at
different points of time as and when the specific subject area requirements are
clear. The data marts are integrated or combined together to form a data
warehouse. Separate data marts are combined through the use of conformed
dimensions and conformed facts. A conformed dimension and a conformed fact is
one that can be shared across data marts.
A Conformed dimension has consistent dimension keys, consistent attribute names
and consistent values across separate data marts. The conformed dimension means
exact same thing with every fact table it is joined.
A Conformed fact has the same definition of measures, same dimensions joined to
it and at the same granularity across data marts.
8
9.
The bottom up approach helps us incrementally build the warehouse by developing
and integrating data marts as and when the requirements are clear. We don’t have
to wait for knowing the overall requirements of the warehouse. We should
implement the bottom up approach when
1. We have initial cost and time constraints.
2. The complete warehouse requirements are not clear. We have clarity to only
one data mart.
Merits of Bottom Up approach:
• It does not require high initial costs and have a faster implementation time;
hence the business can start using the marts much earlier as compared to
the top-down approach.
Drawbacks of Bottom Up approach:
• It stores data in the de normalized format, so there would be high space
usage for detailed data.
• We have a tendency of not keeping detailed data in this approach hence
losing out on advantage of having detail data.
1.2.2 Design considerations
A successful data warehouse designer must adopt a holistic approach by
considering all data warehouse components as parts of a single complex system,
and take into account all possible data sources and all known usage requirements.
Most successful data warehouses have the following common characteristics:
1. Are based on a dimensional model
2. Contain historical and current data
3. Include both detailed and summarized data
4. Consolidate disparate data from multiple sources while retaining
consistency
Data warehouse is difficult to build due to the following reason:
• Heterogeneity of data sources
• Use of historical data
• Growing nature of data base
Data warehouse design approach muse be business driven, continuous and iterative
engineering approach. In addition to the general considerations there are following
specific points relevant to the data warehouse design:
1. Data Content
The content and structure of the data warehouse are reflected in its data model.
The data model is the template that describes how information will be organized
within the integrated warehouse framework. The data in a data warehouse must be
9
10.
a detailed data. It must be formatted, cleaned up and transformed to fit the
warehouse data model.
2. Meta Data
It defines the location and contents of data in the warehouse. Meta data is
searchable by users to find definitions or subject areas. In other words, it must
provide decision support oriented pointers to warehouse data and thus provides a
logical link between warehouse data and decision support applications.
3. Data Distribution
One of the biggest challenges when designing a data warehouse is the data
placement and distribution strategy. Data volumes continue to grow in nature.
Therefore, it becomes necessary to know how the data should be divided across
multiple servers and which users should get access to which types of data. The
data can be distributed based on the subject area, location (geographical region), or
time (current, month, year).
4. Tools
A number of tools are available that are specifically designed to help in the
implementation of the data warehouse. All selected tools must be compatible with
the given data warehouse environment and with each other. All tools must be able
to use a common Meta data repository.
Design steps
The following nine-step method is followed in the design of a data warehouse:
1. Choosing the subject matter
2. Deciding what a fact table represents
3. Identifying and conforming the dimensions
4. Choosing the facts
5. Storing pre calculations in the fact table
6. Rounding out the dimension table
7. Choosing the duration of the db
8. The need to track slowly changing dimensions
9. Deciding the query priorities and query models
1.2.3 Technical Considerations
A number of technical issues are to be considered when designing a data
warehouse environment. These issues include:
• The hardware platform that would house the data warehouse
• The DBMS that supports the warehouse database
• The communication infrastructure that connects data marts, operational
systems and end users
• The hardware and software to support meta data repository
• The systems management framework that enables centralized management
and administration of the entire environment.
10
11.
1.2.4 Implementation Considerations
The following logical steps needed to implement a data warehouse:
• Collect and analyze business requirements
• Create a data model and a physical design
• Define data sources
• Choose the database technology and platform
• Extract the data from operational database, transform it, clean it up and load
it into the warehouse
• Choose database access and reporting tools
• Choose database connectivity software
• Choose data analysis and presentation software
• Update the data warehouse
Access Tools
Data warehouse implementation relies on selecting suitable data access tools. The
best way to choose this is based on the type of data and the kind of access it
permits for a particular user. The following lists the various types of data that can
be accessed:
• Simple tabular form data
• Ranking data
• Multivariable data
• Time series data
• Graphing, charting and pivoting data
• Complex textual search data
• Statistical analysis data
• Data for testing of hypothesis, trends and patterns
• Predefined repeatable queries
• Ad hoc user specified queries
• Reporting and analysis data
• Complex queries with multiple joins, multi level sub queries and
sophisticated search criteria
Data Extraction, Clean Up, Transformation and Migration
A proper attention must be paid to data extraction which represents a success
factor for data warehouse architecture. When implementing data warehouse the
following selection criteria should be considered:
• Timeliness of data delivery to the warehouse
• The tool must have the ability to identify the particular data and that can be
read by conversion tool
11
12.
• The tool must support flat files, indexed files since corporate data is still in
this type
• The tool must have the capability to merge data from multiple data stores
• The tool should have specification interface to indicate the data to be
extracted
• The tool should have the ability to read data from data dictionary
• The code generated by the tool should be completely maintainable
• The tool should permit the user to extract the required data
• The tool must have the facility to perform data type and character set
translation
• The tool must have the capability to create summarization, aggregation and
derivation of records
• The data warehouse database system must be able to perform loading data
directly from these tools
Data Placement Strategies
As a data warehouse grows, there are at least two options for data placement. One
is to put some of the data in the data warehouse into another storage media. The
second option is to distribute the data in the data warehouse across multiple
servers.
It considers Data Replication and Database gateways.
Metadata
Meta data can define all data elements and their attributes, data sources and timing
and the rules that govern data use and data transformations.
User Sophistication Levels
The users of data warehouse data can be classified on the basis of their skill level
in accessing the warehouse. There are three classes of users:
Casual users: are most comfortable in retrieving information from warehouse in
pre defined formats and running pre existing queries and reports. These users do
not need tools that allow for building standard and ad hoc reports
Power Users: can use pre defined as well as user defined queries to create simple
and ad hoc reports. These users can engage in drill down operations. These users
may have the experience of using reporting and query tools.
Expert users: These users tend to create their own complex queries and perform
standard analysis on the info they retrieve. These users have the knowledge about
the use of query and report tools.
1.2 Multi-Tier Architecture
The functions of data warehouse are based on the relational data base technology.
The relational data base technology is implemented in parallel manner.
12
13.
There are two advantages of having parallel relational data base technology for
data warehouse:
• Linear Speed up: refers the ability to increase the number of processor to
reduce response time
• Linear Scale up: refers the ability to provide same performance on the same
requests as the database size increases
1.3.1 Types of parallelism
There are two types of parallelism:
• Inter Query Parallelism: In which different server threads or processes
handle multiple requests at the same time.
• Intra Query Parallelism: This form of parallelism decomposes the serial
SQL query into lower level operations such as scan, join, sort etc. Then
these lower level operations are executed concurrently in parallel.
Intra query parallelism can be done in either of two ways:
• Horizontal Parallelism: the data base is partitioned across multiple disks
and parallel processing occurs within a specific task that is performed
concurrently on different processors against different set of data.
• Vertical Parallelism: This occurs among different tasks. All query
components such as scan, join, sort etc are executed in parallel in a
pipelined fashion. In other words, an output from one task becomes an input
into another task.
1.3.2 Database Architecture
There are three DBMS software architecture styles for parallel processing:
1. Shared memory or shared everything Architecture
2. Shared disk architecture
3. Shared nothing architecture
1. Shared Memory Architecture
Tightly coupled shared memory systems have the following characteristics:
• Multiple Processor Units share memory.
• Each Processor Unit has full access to all shared memory through a
common bus.
• Communication between nodes occurs via shared memory.
• Performance is limited by the bandwidth of the memory bus.
13
Interconnection Network
Process
or Unit
(PU)
Process
or Unit
(PU)
Process
or Unit
(PU)
Processo
r
Unit
Global Shared Memory
14.
Fig. 1.3.2.1 Shared Memory Architecture
Symmetric multiprocessor (SMP) machines are often nodes in a cluster. Multiple
SMP nodes can be used with Oracle Parallel Server in a tightly coupled system,
where memory is shared by the multiple Processor Units, and is accessible by all
the Processor Units through a memory bus. Examples of tightly coupled systems
include the Pyramid, Sequent, and Sun SparcServer.
Performance is limited in a tightly coupled system by the factors:
• Memory bandwidth
• Processor Unit to Processor Unit communication bandwidth
• Memory availability
• I/O bandwidth and
• Bandwidth of the common bus.
Parallel processing advantages of shared memory systems are these:
• Memory access is cheaper than inter-node communication. This means that
internal synchronization is faster than using the Lock Manager.
• Shared memory systems are easier to administer than a cluster.
A disadvantage of shared memory systems for parallel processing is as follows:
• Scalability is limited by bus bandwidth and latency, and by available
memory.
2. Shared Disk Architecture
Shared disk systems are typically loosely coupled. Such systems, illustrated
in following figure, have the following characteristics:
• Each node consists of one or more Processor Units and associated memory.
• Memory is not shared between nodes.
• Communication occurs over a common high-speed bus.
• Each node has access to the same disks and other resources.
• A node can be an SMP if the hardware supports it.
• Bandwidth of the high-speed bus limits the number of nodes of the system.
Fig. 1.3.2.2 Shared Disk Architecture
14
Interconnection Network
Processo
r
Unit
(PU)
Processo
r
Unit
(PU)
Processo
r
Unit
(PU)
Processor
Unit (PU)
Global Shared Memory
15.
Each node is having its own data cache as the memory is not shared among the
nodes. Cache consistency must be maintained across the nodes and a lock manager
is needed to maintain the consistency. Additionally, instance locks using the DLM
on the Oracle level must be maintained to ensure that all nodes in the cluster see
identical data.
There is additional overhead in maintaining the locks and ensuring that the data
caches are consistent. The performance impact is dependent on the hardware and
software components, such as the bandwidth of the high-speed bus through which
the nodes communicate, and DLM performance.
Merits of shared disk systems:
• Shared disk systems permit high availability. All data is accessible even if
one node dies.
• These systems have the concept of one database, which is an advantage
over shared nothing systems.
• Shared disk systems provide for incremental growth.
Drawbacks of shared disk systems:
• Inter-node synchronization is required, involving DLM overhead and
greater dependency on high-speed interconnect.
• If the workload is not partitioned well, there may be high synchronization
overhead.
• There is operating system overhead of running shared disk software.
3. Shared Nothing Architecture
Shared nothing systems are typically loosely coupled. In shared nothing systems
only one CPU is connected to a given disk. If a table or database is located on that
disk, access depends entirely on the Processor Unit which owns it. Shared nothing
systems can be represented as follows:
Fig. 1.3.2.3 Distributed Memory Architecture
Shared nothing systems are concerned with access to disks, not access to memory.
Nonetheless, adding more PUs and disks can improve scaleup. Oracle Parallel
Server can access the disks on a shared nothing system as long as the operating
15
Interconnection Network
Processo
r
Unit
(PU)
Processo
r
Unit
(PU)
Processo
r
Unit
(PU)
Processor
Unit (PU)
Local
Memory
Local
Memory
Local
Memory
Local
Memory
16.
system provides transparent disk access, but this access is expensive in terms of
latency.
Advantages of Shared nothing systems:
• Shared nothing systems provide for incremental growth.
• System growth is practically unlimited.
• MPPs are good for read-only databases and decision support applications.
• Failure is local: if one node fails, the others stay up.
Drawbacks of Shared nothing systems:
• More coordination is required.
• More overhead is required for a process working on a disk belonging to
another node.
• If there is a heavy workload of updates or inserts, as in an online transaction
processing system, it may be worthwhile to consider data-dependent routing
to alleviate contention.
1.3 Data Warehousing Schema
There are three basic schemas that are used in dimensional modeling:
1. Star schema
2. Snowflake schema
3. Fact constellation schema
1.4.1 Star schema
The multidimensional view of data that is expressed using relational data base
semantics is provided by the data base schema design called star schema. The
basic of star schema is that information can be classified into two groups:
• Facts
• Dimension
Star schema has one large central table (fact table) and a set of smaller tables
(dimensions) arranged in a radial pattern around the central table.
• Facts are core data element being analyzed
• Dimensions are attributes about the facts.
The determination of which schema model should be used for a data warehouse
should be based upon the analysis of project requirements, accessible tools and
project team preferences.
16
17.
Fig. 1.4.1.1 Star Schema
Star schema has points radiating from a center. The center of the star consists of
fact table and the points of the star are the dimension tables. Usually the fact tables
in a star schema are in third normal form (3NF) whereas dimensional tables are de-
normalized. Star schema is the simplest architecture and is most commonly used
and recommended by Oracle.
Fact Tables
A fact table is a table that contains summarized numerical and historical data
(facts) and a multipart index composed of foreign keys from the primary keys of
related dimension tables.
A fact table typically has two types of columns: foreign keys to dimension tables
and measures those that contain numeric facts. A fact table can contain fact's data
on detail or aggregated level.
Dimension Tables
Dimensions are categories by which summarized data can be viewed. E.g. a profit
summary in a fact table can be viewed by a Time dimension (profit by month,
quarter, year), Region dimension (profit by country, state, city), Product dimension
(profit for product1, product2).
A dimension is a structure usually composed of one or more hierarchies that
categorizes data. If a dimension hasn't got a hierarchies and levels it is called flat
dimension or list. The primary keys of each of the dimension tables are part of the
composite primary key of the fact table. Dimensional attributes help to describe
the dimensional value. They are normally descriptive, textual values. Dimension
tables are generally small in size then fact table.
Measures
Measures are numeric data based on columns in a fact table. They are the primary
data which end users are interested in. E.g. a sales fact table may contain a profit
measure which represents profit on each sale.
17
18.
Cubes are data processing units composed of fact tables and dimensions from the
data warehouse. They provide multidimensional views of data, querying and
analytical capabilities to clients.
The main characteristics of star schema:
• Simple structure and easy to understand.
• Great query effectives for small number of tables to join
• Relatively long time of loading data into dimension tables for de-
normalization, redundancy data caused that size of the table could be
large.
• The most commonly used in the data warehouse implementations.
1.4.2 Snowflake schema: is the result of decomposing one or more of the
dimensions. The many-to-one relationships among sets of attributes of a dimension
can separate new dimension tables, forming a hierarchy. The decomposed
snowflake structure visualizes the hierarchical structure of dimensions very well.
1.4.3 Fact constellation schema: For each star schema it is possible to construct
fact constellation schema. The fact constellation architecture contains multiple fact
tables that share many dimension tables.
The main shortcoming of the fact constellation schema is a more complicated
design because many variants for particular kinds of aggregation must be
considered and selected.
1.4 Multidimensional data model
Multidimensional data model is to view it as a cube. The cable at the left contains
detailed sales data by product, market and time. The cube on the right associates
sales number (unit sold) with dimensions-product type, market and time with the
unit variables organized as cell in an array.
This cube can be expended to include another array-price-which can be associates
with all or only some dimensions. As number of dimensions increases number of
cubes cell increase exponentially.
Dimensions are hierarchical in nature i.e. time dimension may contain hierarchies
for years, quarters, months, week and day. GEOGRAPHY may contain country,
state, city etc.
18
19.
Fig. 1.5.1 Multidimensional cube
Each side of the cube represents one of the elements of the question. The x-axis
represents the time, the y-axis represents the products and the z-axis represents
different centers. The cells in the cube represents the number of product sold or
can represent the price of the items.
When the size of the dimension increases, the size of the cube will also increase
exponentially. The time response of the cube depends on the size of the cube.
1.5.1 Operations in Multidimensional Data Model:
• Aggregation (roll-up)
– dimension reduction: e.g., total sales by city
– summarization over aggregate hierarchy: e.g., total sales by city and
year -> total sales by region and by year
• Selection (slice) defines a sub cube
– e.g., sales where city = Palo Alto and date = 1/15/96
• Navigation to detailed data (drill-down)
– e.g., (sales - expense) by city, top 3% of cities by average income
• Visualization Operations (e.g., Pivot or dice)
1.5 OLAP operations
OLAP stands for Online Analytical Processing. It uses database tables (fact and
dimension tables) to enable multidimensional viewing, analysis and querying of
large amounts of data. OLAP technology could provide management with fast
answers to complex queries on their operational data or enable them to analyze
their company's historical data for trends and patterns.
Online Analytical Processing (OLAP) applications and tools are those that are
designed to ask “complex queries of large multidimensional collections of data.”
Operations:
Roll up (drill-up): summarize data
19
20.
by climbing up hierarchy or by dimension reduction
Drill down (roll down): reverse of roll-up
from higher level summary to lower level summary or detailed data,
or introducing new dimensions
Slice and dice:
project and select
Pivot (rotate):
reorient the cube, visualization, 3D to series of 2D planes.
Other operations
drill across: involving (across) more than one fact table
drill through: through the bottom level of the cube to its back-end
relational tables (using SQL)
1.6.1 OLAP Guidelines
Dr. E.F. Codd the “father” of the relational model, created a list of rules to deal
with the OLAP systems. Users should priorities these rules according to their
needs to match their business requirements. These rules are:
1) Multidimensional conceptual view: The OLAP should provide an
appropriate multidimensional Business model that suits the Business
problems and Requirements.
2) Transparency: The OLAP tool should provide transparency to the input data
for the users.
3) Accessibility: The OLAP tool should only access the data required only to
the analysis needed.
4) Consistent reporting performance: The Size of the database should not
affect in any way the performance.
5) Client/server architecture: The OLAP tool should use the client server
architecture to ensure better performance and flexibility.
6) Generic dimensionality: Data entered should be equivalent to the structure
and operation requirements.
7) Dynamic sparse matrix handling: The OLAP too should be able to manage
the sparse matrix and so maintain the level of performance.
8) Multi-user support: The OLAP should allow several users working
concurrently to work together.
9) Unrestricted cross-dimensional operations: The OLAP tool should be able
to perform operations across the dimensions of the cube.
10)Intuitive data manipulation. “Consolidation path re-orientation, drilling
down across columns or rows, zooming out, and other manipulation
inherent in the consolidation path outlines should be accomplished via
direct action upon the cells of the analytical model, and should neither
require the use of a menu nor multiple trips across the user interface.”
11)Flexible reporting: It is the ability of the tool to present the rows and
column in a manner suitable to be analyzed.
20
21.
12)Unlimited dimensions and aggregation levels: This depends on the kind of
Business, where multiple dimensions and defining hierarchies can be made.
In addition to these guidelines an OLAP system should also support:
• Comprehensive database management tools: This gives the database
management to control distributed Businesses
• The ability to drill down to detail source record level: Which requires that
The OLAP tool should allow smooth transitions in the multidimensional
database.
• Incremental database refresh: The OLAP tool should provide partial refresh.
• Structured Query Language (SQL interface): the OLAP system should be
able to integrate effectively in the surrounding enterprise environment.
1.7Data warehouse implementation
1.7.1 Efficient Data Cube Computation
Data cube can be viewed as a lattice of cuboids
The bottom-most cuboid is the base cuboid
The top-most cuboid (apex) contains only one cell
How many cuboids in an n-dimensional cube with L levels?
Materialization of data cube
Materialize every (cuboid) (full materialization), none (no
materialization), or some (partial materialization)
Selection of which cuboids to materialize
Based on size, sharing, access frequency, etc.
Cube definition and computation in DMQL
define cube sales[item, city, year]: sum(sales_in_dollars)
compute cube sales
Transform it into a SQL-like language (with a new operator cube by,
introduced by Gray et al.’96)
SELECT item, city, year, SUM (amount)
FROM SALES
CUBE BY item, city, year
Need compute the following Group-Bys
(date, product, customer),
(date,product),(date, customer), (product, customer),
(date), (product), (customer)
()
21
22.
Join index: JI(R-id, S-id) where R (R-id, …) >< S (S-id, …)
Traditional indices map the values to a list of record ids
It materializes relational join in JI file and speeds up relational join
— a rather costly operation
In data warehouses, join index relates the values of the dimensions of a start
schema to rows in the fact table.
E.g. fact table: Sales and two dimensions city and product
A join index on city maintains for each distinct city a list of R-
IDs of the tuples recording the Sales in the city
Join indices can span multiple dimensions
22
23.
1.8 Data Warehouse to Data Mining
1.8.1 Data Warehouse Usage
Three kinds of data warehouse applications
Information processing
supports querying, basic statistical analysis, and reporting
using crosstabs, tables, charts and graphs
Analytical processing
multidimensional analysis of data warehouse data
supports basic OLAP operations, slice-dice, drilling, pivoting
Data mining
knowledge discovery from hidden patterns
supports associations, constructing analytical models,
performing classification and prediction, and presenting the
mining results using visualization tools.
Differences among the three tasks
23
24.
1.8.2 From On-Line Analytical Processing to On Line Analytical
Mining (OLAM
Why online analytical mining?
High quality of data in data warehouses
DW contains integrated, consistent, cleaned data
Available information processing structure surrounding data
warehouses
ODBC, OLEDB, Web accessing, service facilities, reporting
and OLAP tools
OLAP-based exploratory data analysis
mining with drilling, dicing, pivoting, etc.
On-line selection of data mining functions
integration and swapping of multiple mining functions,
algorithms, and tasks.
1.8.3Architecture of OLAM
24
25.
UNIT II DATA PREPROCESSING, LANGUAGE, ARCHITECTURES,
CONCEPT DESCRIPTION
Why preprocessing − Cleaning − Integration − Transformation − Reduction −
Discretization – Concept hierarchy generation − Data mining primitives − Query
language − Graphical user interfaces − Architectures − Concept description − Data
generalization − Characterizations − Class comparisons − Descriptive statistical
measures.
2.1 Data preprocessing
Data preprocessing transforms the data into a format that will be more easily and
effectively processed for the purpose of the user.
Data preprocessing describes any type of processing performed on raw data to
prepare it for another processing procedure. Commonly used as a preliminary data
mining practice, data preprocessing transforms the data into a format that will be
more easily and effectively processed for the purpose of the user.
We need data processing as data in the real world are dirty. It can be in incomplete,
noisy and inconsistent from. These data needs to be preprocessed in order to
improve the quality of the data, and quality of the mining results.
• If no quality data, then no quality mining results. The quality decision is always
based on the quality data.
• If there is much irrelevant and redundant information present or noisy and
unreliable data, then knowledge discovery during the training phase is more
difficult.
• Incomplete data may come from
o “Not applicable” data value when collected
o Different considerations between the time when the data was collected
and when it is analyzed.
o Due to Human/hardware/software problems
o e.g., occupation=“ ”.
• Noisy data (incorrect values) may come from
o Faulty data collection by instruments
o Human or computer error at data entry
o Errors in data transmission and contain errors or outliers data. e.g.,
Salary=“-10”
• Inconsistent data may come from
o Different data sources
o Functional dependency violation (e.g., modify some linked data)
o Having discrepancies in codes or names. e.g., Age=“42”
Birthday=“03/07/1997”
2.5.1 Major Tasks in Data Preprocessing
• Data cleaning
o Fill in missing values, smooth noisy data, identify or remove outliers,
and resolve inconsistencies
25
26.
• Data integration
o Integration of multiple databases, data cubes, or files
• Data transformation
o Normalization and aggregation
• Data reduction
o Obtains reduced representation in volume but produces the same or
similar analytical results
• Data discretization
o Part of data reduction but with particular importance, especially for
numerical data
Fig. 2.5.1.1 Forms of Data Preprocessing
2.2 Data cleaning:
Data cleaning routines attempt to fill in missing values, smooth out noise while
identifying outliers, and correct inconsistencies in the data.
i. Missing Values:
The various methods for handling the problem of missing values in data tuples
include:
(a) Ignoring the tuple: When the class label is missing the tuple can be
ignored. This method is not very effective unless the tuple contains several
attributes with missing values. It is especially poor when the percentage of
missing values per attribute varies considerably.
(b) Manually filling in the missing value: In general, this approach is
time-consuming and may not be a reasonable task for large data sets with
26
27.
many missing values, especially when the value to be filled in is not easily
determined.
(c) Using a global constant to fill in the missing value: Replace all
missing attribute values by the same constant, such as a label like
“Unknown,” or −∞. If missing values are replaced by, say, “Unknown,”
then the mining program may mistakenly think that they form an interesting
concept, since they all have a value in common — that of “Unknown” .
(d) Using the attribute mean for quantitative (numeric) values or
attribute mode for categorical (nominal) values, for all samples
belonging to the same class as the given tuple: For example, if classifying
customers according to credit risk, replace the missing value with the
average income value for customers in the same credit risk category as that
of the given tuple.
(e) Using the most probable value to fill in the missing value: This may
be determined with regression, inference-based tools using Bayesian
formalism, or decision tree induction. For example, using the other
customer attributes in your data set, you may construct a decision tree to
predict the missing values for income.
ii. Noisy data:
Noise is a random error or variance in a measured variable. Data smoothing tech is
used for removing such noisy data.
Several Data smoothing techniques used:
a. Binning Method
b. Regression Method
c. Cluster Method
1 Binning methods: Binning methods smooth a sorted data value by consulting
the neighborhood", or values around it. The sorted values are distributed into a
number of 'buckets', or bins. Because binning methods consult the neighborhood of
values, they perform local smoothing.
In this technique,
1. The data for first sorted
2. Then the sorted list partitioned into equi-depth of bins.
3. Then one can smooth by bin means, smooth by bin median, smooth by bin
boundaries, etc.
a. Smoothing by bin means: Each value in the bin is replaced by the
mean value of the bin.
b. Smoothing by bin medians: Each value in the bin is replaced by the
bin median.
c. Smoothing by boundaries: The min and max values of a bin are
identified as the bin boundaries. Each bin value is replaced by the
closest boundary value.
• Example: Binning Methods for Data Smoothing
27
28.
o Sorted data for price (in dollars): 4, 8, 9, 15, 21, 21, 24, 25, 26, 28,
29, 34
o Partition into (equi-depth) bins(equi depth of 3 since each bin
contains three values):
Bin 1: 4, 8, 9, 15
Bin 2: 21, 21, 24, 25
Bin 3: 26, 28, 29, 34
o Smoothing by bin means:
Bin 1: 9, 9, 9, 9
Bin 2: 23, 23, 23, 23
Bin 3: 29, 29, 29, 29
o Smoothing by bin boundaries:
Bin 1: 4, 4, 4, 15
Bin 2: 21, 21, 25, 25
Bin 3: 26, 26, 26, 34
In smoothing by bin means, each value in a bin is replaced by the mean value of
the bin. For example, the mean of the values 4, 8, and 15 in Bin 1 is 9. Therefore,
each original value in this bin is replaced by the value 9.
Smoothing by bin medians can be employed, in which each bin value is replaced
by the bin median. In smoothing by bin boundaries, the minimum and maximum
values in a given bin are identified as the bin boundaries. Each bin value is then
replaced by the closest boundary value.
Suppose that the data for analysis include the attribute age. The age values for the
data tuples are (in increasing order): 13, 15, 16, 16, 19, 20, 20, 21, 22, 22, 25, 25,
25, 25, 30, 33, 33, 35, 35, 35, 35, 36, 40, 45, 46, 52, 70.
(a) Use smoothing by bin means to smooth the above data, using a bin depth of 3.
The following steps are required to smooth the above data using smoothing by bin
means with a bin depth of 3.
• Step 1: Sort the data. (This step is not required here as the data are already
sorted.)
• Step 2: Partition the data into equidepth bins of depth 3.
Bin 1: 13, 15, 16 Bin 2: 16, 19, 20 Bin 3: 20, 21, 22
Bin 4: 22, 25, 25 Bin 5: 25, 25, 30 Bin 6: 33, 33, 35
Bin 7: 35, 35, 35 Bin 8: 36, 40, 45 Bin 9: 46, 52, 70
• Step 3: Calculate the arithmetic mean of each bin.
• Step 4: Replace each of the values in each bin by the arithmetic mean
calculated for the bin.
Bin 1: 14, 14, 14 Bin 2: 18, 18, 18 Bin 3: 21, 21, 21
Bin 4: 24, 24, 24 Bin 5: 26, 26, 26 Bin 6: 33, 33, 33
28
29.
Bin 7: 35, 35, 35 Bin 8: 40, 40, 40 Bin 9: 56, 56, 56
2 Regression: smooth by fitting the data into regression functions.
• Linear regression involves finding the best of line to fit two variables, so that
one variable can be used to predict the other.
Fig. 2.5.1.2 Regression
• Multiple linear regression is an extension of linear regression, where more than
two variables are involved and the data are fit to a multidimensional surface.
Using regression to find a mathematical equation to fit the data helps smooth out
the noise.
3. Clustering: Outliers in the data may be detected by clustering, where similar
values are organized into groups, or ‘clusters’. Values that fall outside of the set of
clusters may be considered outliers.
Fig. 2.5.1.3 Clustering
iii. Data Cleaning Process:
• Field overloading: is a kind of source of errors that typically occurs when
developers compress new attribute definitions into unused portions of already
defined attributes.
• Unique rule is a rule says that each value of the given attribute must be
different from all other values of that attribute
• Consecutive rule is a rule says that there can be no missing values between the
lowest and highest values of the attribute and that all values must also be
unique.
• Null rule specifies the use of blanks, question marks, special characters or
other strings that may indicate the null condition and how such values should
be handled.
2.3 Data Integration
It combines data from multiple sources into a coherent store. There are number of
issues to consider during data integration.
Issues:
• Schema integration: refers integration of metadata from different sources.
29
30.
• Entity identification problem: Identifying entity in one data source similar
to entity in another table. For example, customer_id in one database and
customer_no in another database refer to the same entity
• Detecting and resolving data value conflicts: Attribute values from
different sources can be different due to different representations, different
scales. E.g. metric vs. British units
• Redundancy: Redundancy can occur due to the following reasons:
• Object identification: The same attribute may have different names
in different db
• Derived Data: one attribute may be derived from another attribute.
• Correlation analysis is used to detect the redundancy.
2.4 Data Transformation
In data transformation, the data are transformed or consolidated into forms
appropriate for mining.
Data transformation can involve the following:
• Smoothing is used to remove noise from the data. It includes binning,
regression, and clustering.
• Aggregation operations such as are applied to the data.
o For example, the daily sales data may be aggregated so as to compute
monthly and annual total amounts.
• Generalization of the data, where low-level or “primitive” (raw) data are
replaced by higher-level concepts through the use of concept hierarchies. For
example, categorical attributes, like street, can be generalized to higher-level
concepts, like city or country.
• Normalization is used to scale the attribute data to fall within a small specified
range, such as -1:0 to 1:0, or 0:0 to 1:0.
• Attribute construction (or feature construction) is use to construct new
attributes which can be added from the given set of attributes to help the
mining process.
2.5 Data Reduction
Data reduction is a technique used to have a reduced representation of data set.
Various Strategies used for data reduction:
1. Data cube aggregation uses aggregation operations that can be applied to the
data in the construction of a data cube.
2. Attribute subset selection, where irrelevant, weakly relevant or redundant
attributes or dimensions may be detected and removed.
3. Dimensionality reduction, where encoding mechanisms are used to reduce the
data set size.
30
31.
4. Numerosity reduction, where the data are replaced or estimated by smaller data
representations such as parametric models or nonparametric methods such as
clustering, sampling, and the use of histograms.
2.6.Data Discretization
Raw data values for attributes are replaced by ranges or higher conceptual levels in
data discretization.
The various methods used in Data Discretization are Binning, Histogram Analysis,
Entropy-Based Discretization, Interval merging by x2
analysis and Clustering.
Three types of attributes:
Nominal — values from an unordered set
Ordinal — values from an ordered set
Continuous — real numbers
Discretization:
divide the range of a continuous attribute into intervals
Some classification algorithms only accept categorical attributes.
Reduce data size by discretization.
Concept hierarchies
reduce the data by collecting and replacing low level concepts (such
as numeric values for the attribute age) by higher level concepts
(such as young, middle-aged, or senior).
Prepare for further analysis
Binning
Histogram analysis
Clustering analysis
Entropy-based discretization
Segmentation by natural partitioning
Entropy-Based Discretization
Given a set of samples S, if S is partitioned into two intervals S1 and S2
using boundary T, the entropy after partitioning is
The boundary that minimizes the entropy function over all possible
boundaries is selected as a binary discretization.
The process is recursively applied to partitions obtained until some stopping
criterion is met, e.g.,
Experiments show that it may reduce data size and improve classification
accuracy
31
32.
Concept eneration for Categorical data
Specification of a partial ordering of attributes explicitly at the schema level
by users or experts
Specification of a portion of a hierarchy by explicit data grouping
Specification of a set of attributes, but not of their partial ordering
Specification of only a partial set of attributes
2.7 Data Mining Primitives
Finding all the patterns autonomously in a database? — unrealistic because
the patterns could be too many but uninteresting
Data mining should be an interactive process
User directs what to be mined
Users must be provided with a set of primitives to be used to communicate
with the data mining system
Incorporating these primitives in a data mining query language
More flexible user interaction
Foundation for design of graphical user interface
Standardization of data mining industry and practice
Database or data warehouse name
Database tables or data warehouse cubes
Condition for data selection
Relevant attributes or dimensions
Data grouping criteria
Types of knowledge to be mined
Characterization
Discrimination
Association
Classification/prediction
Clustering
Outlier analysis
Other data mining tasks
Background knowledge:Concept Hierarchies
Schema hierarchy
E.g., street < city < province_or_state < country
Set-grouping hierarchy
E.g., {20-39} = young, {40-59} = middle_aged
32
33.
Operation-derived hierarchy
email address: login-name < department < university < country
Rule-based hierarchy
low_profit_margin (X) <= price(X, P1) and cost (X, P2) and (P1 -
P2) < $50
Measurements of Pattern Interestingness
Simplicity
e.g., (association) rule length, (decision) tree size
Certainty
e.g., confidence, P(A|B) = n(A and B)/ n (B), classification reliability or
accuracy, certainty factor, rule strength, rule quality, discriminating weight, etc.
Utility
potential usefulness, e.g., support (association), noise threshold
(description)
Novelty
not previously known, surprising (used to remove redundant rules, e.g.,
Canada vs. Vancouver rule implication support ratio.
2.8 Data Mining Query Language (DMQL)
Motivation
A DMQL can provide the ability to support ad-hoc and interactive
data mining
By providing a standardized language like SQL
Hope to achieve a similar effect like that SQL has on
relational database
Foundation for system development and evolution
Facilitate information exchange, technology transfer,
commercialization and wide acceptance
Design
DMQL is designed with the primitives described earlier
Syntax for DMQL
Syntax for specification of
task-relevant data
the kind of knowledge to be mined
concept hierarchy specification
interestingness measure
pattern presentation and visualization
Putting it all together — a DMQL query
33
34.
Syntax for task-relevant data specification
use database database_name, or use data warehouse data_warehouse_name
from relation(s)/cube(s) [where condition]
in relevance to att_or_dim_list
order by order_list
group by grouping_list
having condition
Syntax for specifying the kind of knowledge to be mined
Characterization
Mine_Knowledge_Specification ::=
mine characteristics [as pattern_name]
analyze measure(s)
Discrimination
Mine_Knowledge_Specification ::=
mine comparison [as pattern_name]
for target_class where target_condition
{versus contrast_class_i where contrast_condition_i}
analyze measure(s)
Association
Mine_Knowledge_Specification ::=
mine associations [as pattern_name]
Classification
Mine_Knowledge_Specification ::=
mine classification [as pattern_name]
analyze classifying_attribute_or_dimension
Prediction
Mine_Knowledge_Specification ::=
mine prediction [as pattern_name]
analyze prediction_attribute_or_dimension
{set {attribute_or_dimension_i= value_i}}
Syntax for concept hierarchy specification
To specify what concept hierarchies to use
use hierarchy <hierarchy> for <attribute_or_dimension>
We use different syntax to define different type of hierarchies
schema hierarchies
define hierarchy time_hierarchy on date as [date,month quarter,year]
34
35.
set-grouping hierarchies
define hierarchy age_hierarchy for age on customer as
level1: {young, middle_aged, senior} < level0: all
level2: {20, ..., 39} < level1: young
level2: {40, ..., 59} < level1: middle_aged
level2: {60, ..., 89} < level1: senior
Syntax for interestingness measure specification
Interestingness measures and thresholds can be specified by the user with
the statement:
with <interest_measure_name> threshold = threshold_value
Example:
with support threshold = 0.05
with confidence threshold = 0.7
2.9 Designing Graphical User Interfaces based on a data
mining query language
What tasks should be considered in the design GUIs based on a data mining
query language?
Data collection and data mining query composition
Presentation of discovered patterns
Hierarchy specification and manipulation
Manipulation of data mining primitives
Interactive multilevel mining
Other miscellaneous information
2.10 Data Mining System Architectures
Coupling data mining system with DB/DW system
No coupling—flat file processing, not recommended
Loose coupling
Fetching data from DB/DW
Semi-tight coupling—enhanced DM performance
Provide efficient implement a few data mining primitives in a
DB/DW system, e.g., sorting, indexing, aggregation,
histogram analysis, multiway join, precomputation of some
stat functions
Tight coupling—A uniform information processing environment
DM is smoothly integrated into a DB/DW system, mining
query is optimized based on mining query, indexing, query
processing methods, etc.
35
36.
2.11Concept Description
Descriptive vs. predictive data mining
Descriptive mining: describes concepts or task-relevant data sets in
concise, summarative, informative, discriminative forms
Predictive mining: Based on data and analysis, constructs models for
the database, and predicts the trend and properties of unknown data
Concept description:
Characterization: provides a concise and succinct summarization of
the given collection of data
Comparison: provides descriptions comparing two or more
collections of data
Concept Description vs. OLAP
Concept description:
can handle complex data types of the attributes and their
aggregations
a more automated process
OLAP:
restricted to a small number of dimension and measure types
user-controlled process
2.12Data Generalization and Summarization-based
Characterization
Data generalization
A process which abstracts a large set of task-relevant data in a
database from a low conceptual levels to higher ones.
Approaches:
Data cube approach(OLAP approach)
Attribute-oriented induction approach
Characterization: Data Cube Approach (without using AO-Induction)
Perform computations and store results in data cubes
Strength
An efficient implementation of data generalization
Computation of various kinds of measures
e.g., count( ), sum( ), average( ), max( )
Generalization and specialization can be performed on a data cube
by roll-up and drill-down
Limitations
36
37.
handle only dimensions of simple nonnumeric data and measures of
simple aggregated numeric values.
Lack of intelligent analysis, can’t tell which dimensions should be
used and what levels should the generalization reach
Attribute-Oriented Induction
Proposed in 1989 (KDD ‘89 workshop)
Not confined to categorical data nor particular measures.
How it is done?
Collect the task-relevant data( initial relation) using a relational
database query
Perform generalization by attribute removal or attribute
generalization.
Apply aggregation by merging identical, generalized tuples and
accumulating their respective counts.
Interactive presentation with users.
Basic Principles of Attribute-Oriented Induction
Data focusing: task-relevant data, including dimensions, and the result is the
initial relation.
Attribute-removal: remove attribute A if there is a large set of distinct
values for A but (1) there is no generalization operator on A, or (2) A’s
higher level concepts are expressed in terms of other attributes.
Attribute-generalization: If there is a large set of distinct values for A, and
there exists a set of generalization operators on A, then select an operator
and generalize A.
Attribute-threshold control: typical 2-8, specified/default.
Generalized relation threshold control: control the final relation/rule size.
Basic Algorithm for Attribute-Oriented Induction
InitialRel: Query processing of task-relevant data, deriving the initial
relation.
PreGen: Based on the analysis of the number of distinct values in each
attribute, determine generalization plan for each attribute: removal? or how
high to generalize?
PrimeGen: Based on the PreGen plan, perform generalization to the right
level to derive a “prime generalized relation”, accumulating the counts.
37
38.
Presentation: User interaction: (1) adjust levels by drilling, (2) pivoting, (3)
mapping into rules, cross tabs, visualization presentations.
Example
DMQL: Describe general characteristics of graduate students in the Big-
University database
use Big_University_DB
mine characteristics as “Science_Students”
in relevance to name, gender, major, birth_place, birth_date, residence, phone#,
gpa
from student
where status in “graduate”
Corresponding SQL statement:
Select name, gender, major, birth_place, birth_date, residence, phone#, gpa
from student
where status in {“Msc”, “MBA”, “PhD” }
Presentation of Generalized Results
Generalized relation:
Relations where some or all attributes are generalized, with counts or
other aggregation values accumulated.
Cross tabulation:
Mapping results into cross tabulation form (similar to contingency
tables).
Visualization techniques:
Pie charts, bar charts, curves, cubes, and other visual forms.
Quantitative characteristic rules:
Mapping generalized result into characteristic rules with quantitative
information associated with it, e.g.,
2.13 Mining Class Comparisons
Comparison: Comparing two or more classes.
Method:
Partition the set of relevant data into the target class and the
contrasting class(es)
Generalize both classes to the same high level concepts
Compare tuples with the same high level descriptions
Present for every tuple its description and two measures:
support - distribution within single class
comparison - distribution between classes
Highlight the tuples with strong discriminant features
Relevance Analysis:
Find attributes (features) which best distinguish different classes.
38
39.
Task
Compare graduate and undergraduate students using discriminant
rule.
DMQL query
use Big_University_DB
mine comparison as “grad_vs_undergrad_students”
in relevance to name, gender, major, birth_place, birth_date, residence,
phone#, gpa
for “graduate_students”
where status in “graduate”
versus “undergraduate_students”
where status in “undergraduate”
analyze count%
from student
Example: Analytical comparison (2)
Given
attributes name, gender, major, birth_place, birth_date, residence,
phone# and gpa
Gen(ai) = concept hierarchies on attributes ai
Ui = attribute analytical thresholds for attributes ai
Ti = attribute generalization thresholds for attributes ai
R = attribute relevance threshold
Example: Analytical comparison (3)
1. Data collection
target and contrasting classes
2. Attribute relevance analysis
remove attributes name, gender, major, phone#
3. Synchronous generalization
controlled by user-specified dimension thresholds
prime target and contrasting class(es) relations/cuboids
Class Description
Quantitative characteristic rule
necessary
Quantitative discriminant rule
sufficient
Quantitative description rule
necessary and sufficient
39
40.
2.14Mining descriptive statistical measures in large databases
2.14.1 Mining Data Dispersion Characteristics
Motivation
To better understand the data: central tendency, variation and spread
Data dispersion characteristics
median, max, min, quantiles, outliers, variance, etc.
Numerical dimensions correspond to sorted intervals
Data dispersion: analyzed with multiple granularities of precision
Boxplot or quantile analysis on sorted intervals
Dispersion analysis on computed measures
Folding measures into numerical dimensions
Boxplot or quantile analysis on the transformed cube
Measuring the Central Tendency
Mean
Weighted arithmetic mean
Median: A holistic measure
Middle value if odd number of values, or average of the middle two
values otherwise
estimated by interpolation
Mode
Value that occurs most frequently in the data
Unimodal, bimodal, trimodal
Empirical formula:mean-mode=3x(mean-mode)
Measuring the Dispersion of Data
Quartiles, outliers and boxplots
Quartiles: Q1 (25th
percentile), Q3 (75th
percentile)
Inter-quartile range: IQR = Q3 –Q1
Five number summary: min, Q1, M,Q3, max
Boxplot: ends of the box are the quartiles, median is marked,
whiskers, and plot outlier individually
Outlier: usually, a value higher/lower than 1.5 x IQR
Variance and standard deviation
Variance s2
: (algebraic, scalable computation)
Standard deviation s is the square root of variance s2
Boxplot Analysis
Five-number summary of a distribution:
Minimum, Q1, M, Q3, Maximum
Boxplot
40
41.
Data is represented with a box
The ends of the box are at the first and third quartiles, i.e., the height
of the box is IRQ
The median is marked by a line within the box
Whiskers: two lines outside the box extend to Minimum and
Maximum
Graphic Displays of Basic Statistical Descriptions
Histogram: (shown before)
Boxplot: (covered before)
Quantile plot: each value xi is paired with fi indicating that approximately
100 fi % of data are ≤ xi
Quantile-quantile (q-q) plot: graphs the quantiles of one univariant
distribution against the corresponding quantiles of another
Scatter plot: each pair of values is a pair of coordinates and plotted as points
in the plane
Loess (local regression) curve: add a smooth curve to a scatter plot to
provide better perception of the pattern of dependence
UNIT III ASSOCIATION RULES
Association rule mining − Single-dimensional boolean association rules from
transactional databases − Multi level association rules from transaction databases
3.1 Association rule mining:
Association rule mining is used for finding frequent patterns, associations,
correlations, or causal structures among sets of items or objects in transaction
databases, relational databases, and other information repositories. It searches for
interesting relationships among items in a given data set.
3.1.1 Market basket analysis: Electronic shops
A motivating example for association rule mining
• Motivation: finding regularities in data
• What products were often purchased together? — Beer and diapers?!
• What are the subsequent purchases after buying a PC?
• What kinds of DNA are sensitive to this new drug?
• Can we automatically classify web documents?
Association rule mining is used for analyzing buying behavior. Frequently
purchased items can be placed in close proximity in order to further encourage the
sale of such items together.
If customers who purchase computers also tend to buy financial management
software at the same time, then placing the hardware display close to the software
display may help to increase the sales of both of these items.
41
42.
Each basket can then be represented by a Boolean vector of values assigned to
these variable. The Boolean vectors can be analyzed for buying patterns which
reflect items that are frequent associated or purchased together. These patterns can
be represented in the form of association rules.
For example, the information that customers who purchase computers also tend to
buy financial management software at the same time is represented in association
Rule.
computer =>financial management software [support = 2%; confidence =
60%]
Example of association rule mining is market basket analysis. This process
analyzes customer buying habits by finding associations between the different
items that customers place in their “shopping baskets”.
Fig. 3.1.1 Market basket analysis
The discovery of such associations can help retailers develop marketing strategies
by gaining insight into which items are frequently purchased together by
customers. For instance, if customers are buying milk, how likely are they to also
buy bread (and what kind of bread) on the same trip to the supermarket? Such
information can lead to increased sales by helping retailers to do selective
marketing and plan their shelf space.
For instance, placing milk and bread within close proximity may further encourage
the sale of these items together within single visits to the store.
3.1.2 Basic Concepts: Frequent Patterns and Association Rules
• Itemset X={x1, …, xk}
• Find all the rules XàY with min confidence and support
• support, s, probability that a transaction contains X∪Y
42
43.
• confidence, c, conditional probability that a transaction having X also
contains Y.
Rule support and confidence are two measures of rule interestingness that were
described
A support of 2% for association Rule means that 2% of all the transactions
under analysis show that computer and financial management software are
purchased together
A confidence of 60% means that 60% of the customers who purchased a
computer also bought the software. Typically, association rules are considered
interesting if they satisfy both a minimum support threshold and a minimum
confidence threshold. Such thresholds can be set by users or domain experts.
Rules that satisfy both a minimum support threshold (min sup) and a
minimum confidence threshold (min conf) are called strong. By convention, we
write min sup and min conf values so as to occur between 0% and 100%,
• A set of items is referred to as an itemset.
• An itemset that contains k items is a k-itemset.
• The set of computer, financial management software is a 2-itemset.
• The occurrence frequency of an itemset is the number of transactions that
contain the itemset. This is also known as the frequency or support count
of the itemset.
• The number of transactions required for the itemset to satisfy minimum
support is referred to as the minimum support count.
Association rule mining - a two-step process:
Step 1: Find all frequent itemsets. By definition, each of these itemsets will
occur at least as frequently as a pre-determined minimum support count.
Step 2: Generate strong association rules from the frequent itemsets. By
definition, these rules must satisfy minimum support and minimum
confidence.
3.1.3 Association rule mining:
Association rules can be classified in various ways, based on the following criteria:
1. Based on the types of values handled in the rule:
• If a rule concerns associations between the presence or absence of items, it
is a Boolean association rule.
• If a rule describes associations between quantitative items or attributes, then
it is a quantitative association rule. In these rules, quantitative values for
items or attributes are partitioned into intervals.
43
44.
age(X; “30 ……39") ^ income(X; “42K ….. 48K") )=>buys( X, high resolution
TV")
2. Based on the dimensions of data involved in the rule:
If the items or attributes in an association rule each reference only one dimension,
then it is a single- dimensional association rule.
The above rule could be rewritten as
buys(X; “computer") => buys(X; financial management software")
The above example is a single-dimensional association rule since it refers to only
one dimension, i.e., buys. If a rule references two or more dimensions, such as the
dimensions buys, time of transaction, and customer category, then it is a
multidimensional association rule.
3. Based on the levels of abstractions involved in the rule set:
Some methods for association rule mining can find rules at differing levels of
abstraction.
For example, suppose that a set of association rules mined included Rule
age(X,”30…..39")) buys(X; “laptop computer")
age(X; “30 …39") ) buys(X; “computer")
In the above said examples the items bought are referenced at different levels of
abstraction. We refer to the rule set mined as consisting of multilevel association
rules. If, instead, the rules within a given set do not reference items or attributes at
different levels of abstraction, then the set contains single-level association rules.
4. Based on the nature of the association involved in the rule:
Association mining can be extended to correlation analysis, where the absence or
presence of correlated items can be identified.
3.2 Mining single-Dimensional Boolean association rules from
Transactional databases
Different methods for mining the simplest form of association rules -
single-dimensional, single-level, Boolean association rules, such as those
discussed for market basket analysis presenting Apriori. It is a basic algorithm for
finding frequent itemsets. It uses a procedure for generating strong association
rules from frequent itemsets.
3.2.1 The Apriori algorithm: Finding frequent itemsets
Apriori is an influential algorithm for mining frequent itemsets for Boolean
association rules. The name of the algorithm is based on the fact that the algorithm
uses prior knowledge of frequent itemset properties.
44
45.
Apriori employs an iterative approach known as a level-wise search, where k-
itemsets are used to explore (k+1)-itemsets. First, the set of frequent 1-itemsets is
found. This set is denoted L1. L1 is used to find L2, the frequent 2-itemsets, which
is used to find L3, and so on, until no more frequent k-itemsets can be found. The
finding of each Lk requires one full scan of the database. To improve the
efficiency of the level-wise generation of frequent itemsets, an important property
called the Apriori property is used to reduce the search space.
The Apriori property. All non-empty subsets of a frequent itemset must also be
frequent.
By definition, if an itemset I does not satisfy the minimum support threshold, s,
then I is not frequent, i.e., P(I) < s. If an item A is added to the itemset I, then the
resulting itemset cannot occur more frequently than I. This property belongs to a
special category of properties called anti-monotone in the sense that if a set cannot
pass a test, all of its supersets will fail the same test as well. It is called anti-
monotone because the property is monotonic in the context of failing a test.
1. The join step: To find Lk, a set of candidate k-itemsets is generated by joining
Lk-1 with itself. This set of candidates is denoted Ck. Let l1 and l2 be itemsets in
Lk_1. The notation li[j] refers to the jth item in li.
By convention, Apriori assumes that items within a transaction or itemset are
sorted in increasing lexicographic order. It also ensures that no duplicates are
generated.
3. The prune step: Ck is a superset of Lk, that is, its members may or may not
be frequent, but all of the frequent k-itemsets are included in Ck. A scan of
the database to determine the count of each candidate in Ck would result in
the determination of Lk. Ck can be huge, and so this could involve heavy
computation.
The Apriori Algorithm
Join Step: Ck is generated by joining Lk-1with itself
Prune Step: Any (k-1)-itemset that is not frequent cannot be a subset of a
frequent k-itemset
Pseudo-code:
Ck: Candidate itemset of size k
Lk : frequent itemset of size k
L1 = {frequent items};
for (k = 1; Lk !=∅; k++) do begin
Ck+1 = candidates generated from Lk;
for each transaction t in database do
45
46.
increment the count of all candidates in Ck+1 that are
contained in t
Lk+1 = candidates in Ck+1 with min_support
end
return ∪k Lk;
To reduce the size of Ck, the Apriori property is used as follows. Any (k-1)-
itemset that is not frequent cannot be a subset of a frequent k-itemset. Hence, if
any (k-1)-subset of a candidate k-itemset is not in Lk-1, then the candidate cannot
be frequent either and so can be removed from Ck. This subset testing can be done
quickly by maintaining a hash tree of all frequent itemsets.
46
47.
Fig. 3.2.1.1 Transactional data for an All Electronics branch
.
Let's look at a concrete example of Apriori, based on the All Electronics
transaction database, D, of There are nine transactions in this database.
1. In the first iteration of the algorithm, each item is a member of the set of
candidate 1-itemsets, C1. The algorithm simply scans all of the transactions in
order to count the number of occurrences of each item.
2. Suppose that the minimum transaction support count required is 2 (i.e., min sup
= 2). The set of frequent 1-itemsets, L1, can then be determined. It consists of the
candidate 1-itemsets having minimum support.
3. To discover the set of frequent 2-itemsets, L2, the algorithm uses L1×L1 to
generate a candidate set of 2-itemsets, C2.
4. Next, the transactions in D are scanned and the support count of each candidate
itemset in C2 is accumulated.
5. The set of frequent 2-itemsets, L2, is then determined, consisting of those
candidate 2-itemsets in C2 having minimum support.
6. The generation of the set of candidate 3-itemsets, C3. Based on the Apriori
property that all subsets of a frequent itemset must also be frequent, we can
determine that the four latter candidates cannot possibly be frequent. We therefore
remove them from C3, thereby saving the effort of unnecessarily obtaining their
counts during the subsequent scan of D to determine L3.
7. The transactions in D are scanned in order to determine L3, consisting of those
candidate 3-itemsets in C3 having minimum support.
8. The algorithm uses L3×L3 to generate a candidate set of 4-itemsets, C4.
Generating association rules from frequent itemsets:
Once the frequent itemsets from transactions in a database D have been found, it is
straightforward to generate strong association rules from them (where strong
association rules satisfy both minimum support and minimum confidence). This
can be done for confidence, where the conditional probability is expressed in terms
of itemset support count:
• support count(A U B) is the number of transactions containing the itemsets
AUB, and
• support count(A) is the number of transactions containing the itemset A.
• Based on this equation, association rules can be generated as follows.
• For each frequent itemset, l, generate all non-empty subsets of l.
• For every non-empty subset s of l, output the rule s → (l-s)"
47
48.
where min_conf is the minimum confidence threshold.
Variations of the Apriori algorithm
Many variations of the Apriori algorithm have been proposed. A number of
these variations are enumerated below. Methods 1 to 6 focus on improving the
efficiency of the original algorithm, while methods 7 and 8 consider transactions
over time.
1. A hash-based technique: Hashing itemset counts.
A hash-based technique can be used to reduce the size of the candidate k-
itemsets, Ck, for k > 1. For example, when scanning each transaction in the
database to generate the frequent 1-itemsets, L1, from the candidate
A 2-itemset whose corresponding bucket count in the hash table is below the
support threshold cannot be frequent and thus should be removed from the
candidate set. Such a hash-based technique may substantially reduce the number of
the candidate k-itemsets examined (especially when k = 2).
2. Transaction reduction: Reducing the number of transactions scanned in future
iterations. A transaction which does not contain any frequent k-itemsets cannot
contain any frequent (k + 1)-itemsets. Therefore, such a transaction can be marked
or removed from further consideration since subsequent scans of the database for j-
itemsets, where j > k, will not require it.
3. Partitioning:
It is used for partitioning the data to find candidate itemsets. A partitioning
technique can be used which requires just two database scans to mine the frequent
itemsets. It consists of two phases.
• In Phase I, the algorithm subdivides the transactions of D into n
non-overlapping partitions. If the minimum support threshold for
transactions in D is min_sup, then the minimum itemset support count for a
partition is min_sup*the number of transactions in that partition.
• For each partition, all frequent itemsets within the partition are found.
These are referred to as local frequent itemsets.
• The procedure employs a special data structure which, for each itemset,
records the TID's of the transactions containing the items in the itemset.
This allows it to find all of the local frequent k-itemsets, for k = 1,2 ……..n
in just one scan of the database.
• The collection of frequent itemsets from all partitions forms a global
candidate itemset with respect to D.
• In Phase II, a second scan of D is conducted in which the actual support of
each candidate is assessed in order to determine the global frequent
48
49.
itemsets. Partition size and the number of partitions are set so that each
partition can fit into main memory and therefore be read only once in each
phase.
4. Sampling:
It is used for Mining on a subset of the given data. The basic idea of the sampling
approach is to pick a random sample S of the given data D, and then search for
frequent itemsets in S instead D.
5. Dynamic itemset counting:
It adds candidate itemsets at different points during a scan. A dynamic itemset
counting technique was proposed in which the database is partitioned into blocks
marked by start points.
In this variation, new candidate itemsets can be added at any start point, unlike in
Apriori, which determines new candidate itemsets only immediately prior to each
complete database scan. The technique is dynamic in that it estimates the support
of all of the itemsets that have been counted so far, adding new candidate itemsets
if all of their subsets are estimated to be frequent. The resulting algorithm requires
two database scans.
5.Calendric market basket analysis: Finding itemsets that are frequent in a set of
user-defined time intervals. Calendric market basket analysis uses transaction time
stamps to define subsets of the given database .
Construct FP-tree from a Transaction DB
Steps:
1. Scan DB once, find frequent 1-itemset (single item pattern)
2. Order frequent items in frequency descending order
3. Scan DB again, construct FP-tree
49
50.
Benefits of the FP-tree Structure
Completeness:
never breaks a long pattern of any transaction
preserves complete information for frequent pattern mining
Compactness
reduce irrelevant information—infrequent items are gone
frequency descending ordering: more frequent items are more likely
to be shared
never be larger than the original database (if not count node-links
and counts)
Example: For Connect-4 DB, compression ratio could be over 100
Mining Frequent Patterns Using FP-tree
General idea (divide-and-conquer)
Recursively grow frequent pattern path using the FP-tree
Method
For each item, construct its conditional pattern-base, and then its
conditional FP-tree
Repeat the process on each newly created conditional FP-tree
Until the resulting FP-tree is empty, or it contains only one path
(single path will generate all the combinations of its sub-paths, each
of which is a frequent pattern)
Major Steps to Mine FP-tree
1) Construct conditional pattern base for each node in the FP-tree
50
51.
2) Construct conditional FP-tree from each conditional pattern-base
3) Recursively mine conditional FP-trees and grow frequent patterns obtained so
far
If the conditional FP-tree contains a single path, simply enumerate all
the patterns
3.3 Mining multilevel association rules from transaction
databases
Multilevel association rules
For many applications, it is difficult to find strong associations among data items
at low or primitive levels of abstraction due to the sparsity of data in
multidimensional space. Strong associations discovered at very high concept levels
may represent common sense knowledge.
Example : Suppose we are given the task-relevant set of transactional data in for
sales at the computer department of an All electronics branch, showing the items
purchased for each transaction TID. A concept hierarchy defines sequence of
mappings from a set of low level concepts to higher level, more general concepts.
Data can be generalized by replacing low level concepts within the
Fig. 3.3.1 Class Hierarchy
data by their higher level concepts, or ancestors, from a concept hierarchy 4. The
concept hierarchy has four levels, referred to as levels 0, 1, 2, and 3. By
convention, levels within a concept hierarchy are numbered from top to bottom,
starting with level 0 at the root node for all (the most general abstraction level).
• Level 1 includes computer, software, printer and computer accessory,
• Level 2 includes home computer, laptop computer, education software,
financial management software, .., and
• Level 3 includes IBM home computer, .., Microsoft educational software,
and so on. Level 3 represents the most specific abstraction level of this
hierarchy.
51
52.
Fig. 2.3.2 Multilevel Mining with Reduced Support
Rules generated from association rule mining with concept hierarchies are called
multiple-level or multilevel association rules, since they consider more than one
concept level.
Approaches to mining multilevel association rules
In general, a top-down strategy is employed, where counts are accumulated
for the calculation of frequent itemsets at each concept level, starting at the
concept level 1 and working towards the lower, more specific concept levels, until
no more frequent itemsets can be found. That is, once all frequent itemsets at
concept level 1 are found, then the frequent itemsets at level 2 are found, and so
on.
For each level, any algorithm for discovering frequent itemsets may be used, such
as Apriori or its variations.
1. Using uniform minimum support for all levels (referred to as uniform
support): The same minimum support threshold is used when mining at each level
of abstraction. For example, a minimum support threshold of 5% is used
throughout (e.g., for mining from “computer" down to “laptop computer"). Both
“computer" and “laptop computer" are found to be frequent, while “home
computer" is not.
When a uniform minimum support threshold is used, the search procedure is
simplified. The method is also simple in that users are required to specify only one
minimum support threshold. An optimization technique can be adopted, based on
the knowledge that an ancestor is a superset of its descendents: the search avoids
examining itemsets containing any item whose ancestors do not have minimum
support.
Fig. 2.7.3.2.1 Multilevel Mining with Uniform Support
52
53.
The uniform support approach is unlikely that items at lower levels of abstraction
will occur as frequently as those at higher levels of abstraction. If the minimum
support threshold is set too high, it could miss several meaningful associations
occurring at low abstraction levels.
If the threshold is set too low, it may generate many uninteresting associations
occurring at high abstraction levels. This provides the motivation for the following
approach.
2. Using reduced minimum support at lower levels (referred to as reduced
support): Each level of abstraction has its own minimum support threshold. The
lower the abstraction level is, the smaller the corresponding threshold is. For
example, the minimum support thresholds for levels 1 and 2 are 5% and 3%,
respectively. In this way, “computer", “laptop computer", and “home computer"
are all considered frequent.
Fig. 2.7.3.2.2 Multilevel Mining with Reduced Support
For mining multiple-level associations with reduced support, there are a number of
alternative search strategies.
These include:
1. Level-By-Level Independent: This is a full breadth search, where no background
knowledge of frequent itemsets is used for pruning. Each node is examined,
regardless of whether or not its parent node is found to be frequent.
2. Level-Cross Filtering By Single Item: An item at the i-th
level is examined if and
only if its parent node at the (i-1)-th
level is frequent.
If a node is frequent, its children will be examined; otherwise, its descendents are
pruned from the search. For example, the descendent nodes of “computer" (i.e.,
“laptop computer" and home computer") are not examined, since “computer" is
not frequent.
3. Level-Cross Filtering By K-Item Set: A k-itemset at the ith
level is examined if
and only if its corresponding parent k-itemset at the (i-1)th
level is frequent.
53
54.
UNIT IV CLASSIFICATION AND CLUSTERING
Classification and prediction − Issues − Decision tree induction − Bayesian
classification – Association rule based − Other classification methods − Prediction
− Classifier accuracy − Cluster analysis – Types of data − Categorization of
methods − Partitioning methods − Outlier analysis.
4.1 Classification vs. Prediction
Classification:
predicts categorical class labels
classifies data (constructs a model) based on the training set and the
values (class labels) in a classifying attribute and uses it in
classifying new data
Prediction:
models continuous-valued functions, i.e., predicts unknown or
missing values
Typical Applications
credit approval
target marketing
medical diagnosis
treatment effectiveness analysis
Classification—A Two-Step Process
Model construction: describing a set of predetermined classes
Each tuple/sample is assumed to belong to a predefined class, as
determined by the class label attribute
The set of tuples used for model construction: training set
The model is represented as classification rules, decision trees, or
mathematical formulae
Model usage: for classifying future or unknown objects
Estimate accuracy of the model
The known label of test sample is compared with the
classified result from the model
Accuracy rate is the percentage of test set samples that are
correctly classified by the model
Test set is independent of training set, otherwise over-fitting
will occur
54
55.
Supervised learning (classification)
Supervision: The training data (observations, measurements, etc.) are
accompanied by labels indicating the class of the observations
New data is classified based on the training set
Unsupervised learning (clustering)
The class labels of training data is unknown
Given a set of measurements, observations, etc. with the aim of
establishing the existence of classes or clusters in the data
4.2 Issues regarding classification and prediction (2):
Evaluating Classification Methods
Predictive accuracy
Speed and scalability
time to construct the model
time to use the model
Robustness
handling noise and missing values
Scalability
efficiency in disk-resident databases
Interpretability:
understanding and insight provded by the model
Goodness of rules
55
56.
decision tree size
compactness of classification rules
4.3 Classification by Decision Tree Induction
• Decision tree
o A decision tree is a flowchart-like tree structure, where each internal
node denotes a test on an attribute.
o Each branch represents an outcome of the test, and each leaf node holds
a class label.
o The topmost node in a tree is the root node.
o Internal nodes are denoted by rectangles, and leaf nodes are denoted by
ovals.
o Some decision tree algorithms produce only binary trees whereas others
can produce non binary trees.
• Decision tree generation consists of two phases
o Tree construction
Attribute selection measures are used to select the attribute that
best partitions the tuples into distinct classes.
o Tree pruning
Tree pruning attempts to identify and remove such branches,
with the goal of improving classification accuracy on unseen
data.
• Use of decision tree: Classifying an unknown sample
o Test the attribute values of the sample against the decision
tree
Algorithm for Decision Tree Induction
• Basic algorithm (a greedy algorithm)
o Tree is constructed in a top-down recursive divide-and-conquer manner
o At start, all the training examples are at the root
o Attributes are categorical (if continuous-valued, they are discretized in
advance)
o Examples are partitioned recursively based on selected attributes
o Test attributes are selected on the basis of a heuristic or statistical
measure (e.g., information gain)
• Conditions for stopping partitioning
o All samples for a given node belong to the same class
o There are no remaining attributes for further partitioning – majority
voting is employed for classifying the leaf
o There are no samples left
Attribute Selection Measure
• Information gain
o All attributes are assumed to be categorical and can be modified for
continuous-valued attributes
56
57.
• Gini index
o All attributes are assumed continuous-valued
o Assume there exist several possible split values for each attribute
o May need other tools, such as clustering, to get the possible split values
o Can be modified for categorical attributes
Information Gain (ID3/C4.5)
• Select the attribute with the highest information gain
• Assume there are two classes, P and N
o Let the set of examples S contain p elements of class P and n elements
of class N
o The amount of information, needed to decide if an arbitrary example in
S belongs to P or N is defined as
Information Gain in Decision Tree Induction :
• Assume that using attribute A a set S will be partitioned into sets {S1, S2 , …,
Sv}
• If Si contains pi examples of P and ni examples of N, the entropy, or the
expected information needed to classify objects in all subtrees Si is
• The encoding information that would be gained by branching on A
4.4 Bayesian Classification:
• Probabilistic Learning: Calculate explicit probabilities for hypothesis,
among the most practical approaches to certain types of learning
problems
• Incremental: Each training example can incrementally increase/decrease
the probability that a hypothesis is correct. Prior knowledge can be
combined with observed data.
• Probabilistic Prediction: Predict multiple hypotheses, weighted by their
probabilities
• Standard: Even when Bayesian methods are computationally intractable,
they can provide a standard of optimal decision making against which
other methods can be measured
Bayesian Theorem
57
np
n
np
n
np
p
np
p
npI
++
−
++
−= 22 loglog),(
)(),()( AEnpIAGain −=
58.
• Given training data D, posteriori probability of a hypothesis h, P(h|
D) follows the Bayes theorem
• MAP (maximum posteriori) hypothesis
• Practical difficulty: It requires initial knowledge of many
probabilities, significant computational cost.
Naïve Bayes Classifier
The naïve Bayesian classifier, or simple Bayesian classifier, works as follows:
o Let D be a training set of tuples and their associated class labels. As
usual, each tuple is represented by an n-dimensional attribute vector,
X = (x1, x2, : : : , xn), depicting n measurements made on the tuple
from n attributes, respectively, A1, A2, : : : , An.
o Suppose that there are m classes, C1, C2, : : : , Cm. Given a tuple, X,
the classifier will predict that X belongs to the class having the
highest posterior probability, conditioned on X.
P(Ci|X) > P(Cj|X) for 1≤ j ≤m; j≠ i:
o The class Ci for which P(CijX) is maximized is called the Maximum
posteriori hypothesis.
o As P(X) is constant for all classes, only P(X|Ci)P(Ci) need be
maximized. If the class prior probabilities are not known, then it is
commonly assumed that the classes are equally likely, that is, P(C1)
= P(C2) = …= P(Cm), and we would therefore maximize P(X|Ci).
Otherwise, we maximize P(X|Ci)P(Ci).
o The attributes are conditionally independent of one another, given
the
class label of the tuple
Rule Based Classification
A set of IF-THEN rules are used in Rule Based Classification.
Using IF-THEN Rules for Classification
Rules are a good way of representing information or bits of knowledge. A rule-
based Classifier uses a set of IF-THEN rules for classification.
An IF-THEN rule is an expression of the form
IF condition THEN conclusion.
An example is rule R1,
R1: IF age = youth AND student = yes THEN buys computer = yes.
58
59.
• The “IF”-part (or left-hand side) of a rule is known as the rule antecedent or
precondition. The “THEN”-part (or right-hand side) is the rule consequent.
R1 can also be written as
R1: (age = youth) ^ (student = yes)) (buys computer = yes).
Name Blood Type Give Birth Can Fly Live in Water Class
human warm yes no no mammals
python cold no no no reptiles
salmon cold no no yes fishes
whale warm yes no yes mammals
frog cold no no sometimes amphibians
komodo cold no no no reptiles
bat warm yes yes no mammals
pigeon warm no yes no birds
cat warm yes no no mammals
leopard shark cold yes no yes fishes
turtle cold no no sometimes reptiles
penguin warm no no sometimes birds
porcupine warm yes no no mammals
eel cold no no yes fishes
salamander cold no no sometimes amphibians
gila monster cold no no no reptiles
platypus warm no no no mammals
owl warm no yes no birds
dolphin warm yes no yes mammals
eagle warm no yes no birds
Fig. If Then Example
R1: (Give Birth = no) ∧ (Can Fly = yes) → Birds
R2: (Give Birth = no) ∧ (Live in Water = yes) → Fishes
R3: (Give Birth = yes) ∧ (Blood Type = warm) → Mammals
R4: (Give Birth = no) ∧ (Can Fly = no) → Reptiles
R5: (Live in Water = sometimes) → Amphibians
Application of Rule-Based Classifier
A rule r covers an instance x if the attributes of the instance satisfy the condition of
the rule
R1: (Give Birth = no) ∧ (Can Fly = yes) → Birds
R2: (Give Birth = no) ∧ (Live in Water = yes) → Fishes
R3: (Give Birth = yes) ∧ (Blood Type = warm) → Mammals
R4: (Give Birth = no) ∧ (Can Fly = no) → Reptiles
R5: (Live in Water = sometimes) → Amphibians
The rule R1 covers a hawk => Bird
The rule R3 covers the grizzly bear => Mammal
Advantages of Rule-Based Classifiers
• As highly expressive as decision trees
59
60.
• Easy to interpret
• Easy to generate
• Can classify new instances rapidly
• Performance comparable to decision trees
Prediction
• (Numerical) prediction is similar to classification
o construct a model
o use model to predict continuous or ordered value for a given input
• Prediction is different from classification
o Classification refers to predict categorical class label
o Prediction models continuous-valued functions
• Major method for prediction: regression
o model the relationship between one or more independent or predictor
variables and a dependent or response variable
• Regression analysis
o Linear and multiple regression
o Non-linear regression
o Other regression methods: generalized linear model, Poisson regression,
log-linear models, regression trees
4.5 Association-Based Classification
Several methods for association-based classification
ARCS: Quantitative association mining and clustering of association
rules (Lent et al’97)
It beats C4.5 in (mainly) scalability and also accuracy
Associative classification: (Liu et al’98)
It mines high support and high confidence rules in the form of
“cond_set => y”, where y is a class label
CAEP (Classification by aggregating emerging patterns) (Dong et
al’99)
Emerging patterns (EPs): the itemsets whose support
increases significantly from one class to another
Mine Eps based on minimum support and growth rate
4.6 Other Classification Methods
k-nearest neighbor classifier
case-based reasoning
Genetic algorithm
Rough set approach
Fuzzy set approaches
Instance-Based Methods
60
61.
Instance-based learning:
Store training examples and delay the processing (“lazy evaluation”)
until a new instance must be classified
Typical approaches
k-nearest neighbor approach
Instances represented as points in a Euclidean space.
Locally weighted regression
Constructs local approximation
Case-based reasoning
Uses symbolic representations and knowledge-based
inference
The k-Nearest Neighbor Algorithm
All instances correspond to points in the n-D space.
The nearest neighbor are defined in terms of Euclidean distance.
The target function could be discrete- or real- valued.
For discrete-valued, the k-NN returns the most common value among the k
training examples nearest to xq.
The k-NN algorithm for continuous-valued target functions
Calculate the mean values of the k nearest neighbors
Discussion on the k-NN Algorithm
Distance-weighted nearest neighbor algorithm
Weight the contribution of each of the k neighbors according to their
distance to the query point xq
giving greater weight to closer neighbors
Similarly, for real-valued target functions
Robust to noisy data by averaging k-nearest neighbors
Curse of dimensionality: distance between neighbors could be dominated
by irrelevant attributes.
To overcome it, axes stretch or elimination of the least relevant
attributes.
Remarks on Lazy vs. Eager Learning
Instance-based learning: lazy evaluation
Decision-tree and Bayesian classification: eager evaluation
Key differences
Lazy method may consider query instance xq when deciding how to
generalize beyond the training data D
61
62.
Eager method cannot since they have already chosen global
approximation when seeing the query
Efficiency: Lazy - less time training but more time predicting
Accuracy
Lazy method effectively uses a richer hypothesis space since it uses
many local linear functions to form its implicit global approximation
to the target function
Eager: must commit to a single hypothesis that covers the entire
instance space
4.7 Prediction
Prediction is similar to classification
First, construct a model
Second, use model to predict unknown value
Major method for prediction is regression
Linear and multiple regression
Non-linear regression
Prediction is different from classification
Classification refers to predict categorical class label
Prediction models continuous-valued functions
Predictive Modeling in Databases
Predictive modeling: Predict data values or construct generalized linear
models based on the database data.
One can only predict value ranges or category distributions
Method outline:
Minimal generalization
Attribute relevance analysis
Generalized linear model construction
Prediction
Determine the major factors which influence the prediction
Data relevance analysis: uncertainty measurement, entropy analysis,
expert judgement, etc.
Multi-level prediction: drill-down and roll-up analysis
Regress Analysis and Log-Linear Models in Prediction
Linear regression: Y = α + β X
Two parameters , α and β specify the line and are to be estimated by
using the data at hand.
62
63.
using the least squares criterion to the known values of Y1, Y2, …,
X1, X2, ….
Multiple regression: Y = b0 + b1 X1 + b2 X2.
Many nonlinear functions can be transformed into the above.
Log-linear models:
The multi-way table of joint probabilities is approximated by a
product of lower-order tables.
Probability: p(a, b, c, d) = αab βacχad δbcd
4.8Classification Accuracy: Estimating Error Rates
Partition: Training-and-testing
use two independent data sets, e.g., training set (2/3), test set(1/3)
used for data set with large number of samples
Cross-validation
divide the data set into k subsamples
use k-1 subsamples as training data and one sub-sample as test data
--- k-fold cross-validation
for data set with moderate size
Bootstrapping (leave-one-out)
for small size data
Boosting and Bagging
Boosting increases classification accuracy
Applicable to decision trees or Bayesian classifier
Learn a series of classifiers, where each classifier in the series pays more
attention to the examples misclassified by its predecessor
Boosting requires only linear time and constant space
Boosting Technique (II) — Algorithm
Assign every example an equal weight 1/N
For t = 1, 2, …, T Do
Obtain a hypothesis (classifier) h(t)
under w(t)
Calculate the error of h(t) and re-weight the examples based on the
error
Normalize w(t+1)
to sum to 1
Output a weighted sum of all the hypothesis, with each hypothesis weighted
according to its accuracy on the training set
4.9 Cluster Analysis:
Finding groups of objects such that the objects in a group will be similar (or
related) to one another and different from (or unrelated to) the objects in other
63
64.
groups is called clustering. A cluster is a collection of data objects that are similar
to one another within the same cluster and are dissimilar to the objects in other
clusters. A cluster of data objects can be treated collectively as one group and so
may be considered as a form of data compression.
Clustering is also called data segmentation in some applications because
clustering partitions large data sets into groups according to their similarity.
Clustering can also be used for outlier detection.
A good clustering method will produce high quality clusters with
• high intra-class similarity
• low inter-class similarity.
The quality of a clustering result depends on both the similarity measure used by
the method and its implementation. The quality of a clustering method is also
measured by its ability to discover some or all of the hidden patterns.
The requirements of clustering in data mining:
• Scalability
o Clustering all the data instead of only on samples
• Ability to deal with different types of attributes
• Numerical, binary, categorical, ordinal, linked, and mixture of these types
• Discovery of clusters with arbitrary shape
• Minimal requirements for domain knowledge to determine input parameters
• Ability to deal with noisy data
• Incremental clustering and insensitivity to the order of input records
• High dimensionality
• Constraint-based clustering
User may give inputs on constraints
Use domain knowledge to determine input parameters
• Interpretability and usability
4.10 Types of Data in Cluster Analysis:
A data set to be clustered contains n objects, which may represent persons,
houses, documents, countries, and so on.
Main memory-based clustering algorithms typically operate on the following two
data structures.
• Data matrix: it is otherwise known as object-by-variable structure. This
represents n objects, such as persons, with p variables (also called
measurements or attributes), such as age, height, weight, gender, and so on.
• The structure is in the form of a relational table, or n-by-p matrix:
64
65.
• Dissimilarity matrix: It is otherwise known as object-by-object structure.
This stores a collection of proximities that are available for all pairs of n
objects. It is often represented by an n-by-n table:
• d (i, j) is the measured difference or dissimilarity between objects i and j.
• d(i, j) is a nonnegative number that is close to 0 when objects i and j are
highly similar or “near” each other, and becomes larger the more they
differ.
3.2.1 Interval-Scaled Variables
Interval-scaled variables are continuous measurements of a roughly linear scale.
Typical examples include weight and height, latitude and longitude coordinate and
weather temperature.
The measurement unit used can affect the clustering analysis. For example,
changing measurement units from meters to inches for height, or from kilograms
to pounds for weight, may lead to a very different clustering structure.
Steps involved in Interval-Scaled Variables:
1. Calculate the mean absolute deviation
2. Calculate the standardized measurement (z-score)
3.2.2 Binary Variables
A binary variable has only two states: 0 or 1, where 0 means that the variable is
absent, and 1 means that it is present.
Given the variable smoker describing a patient,
• 1 indicates that the patient smokes,
• 0 indicates that the patient does not.
We need methods specific to binary data are necessary for computing
dissimilarities.
65
|)|...|||(|1
21 fnffffff
mxmxmxns −++−+−=
f
fif
if s
mx
z
−
=
66.
• Distance measure for symmetric binary variables:
• Distance measure for asymmetric binary variables:
• Jaccard coefficient (similarity measure for asymmetric binary variables):
E.g.
Dissimilarity between Binary Variables- patient record
Distance measures are:
3.2.3 Categorical, Ordinal, and Ratio-Scaled Variables
A categorical variable is a generalization of the binary variable in that it can take
on more than two states. For example, map color is a categorical variable that may
have, say, ﬁve states: red, yellow, green, pink, and blue.
Let the number of states of a categorical variable be M. The states can be denoted
by letters, symbols, or a set of integers, such as 1, 2, … , M. These integers are
used for data handling and do not represent any specific ordering.
A discrete ordinal variable resembles a categorical variable, except that the M
states of the ordinal value are ordered in a meaningful sequence. Subjective
assessments of qualities can be measured using ordinal variable.
For example, professional ranks are often enumerated in a sequential order, such as
assistant, associate, and full for professors.
A ratio-scaled variable makes a positive measurement on a nonlinear scale, such
as an exponential scale, approximately following the formula
Where A and B are positive constants, and t typically represents time. Common
examples include the growth of a bacteria population or the decay of a radioactive
element.
3.2.4 Variables of Mixed Types
A database can contain all of the six variable types.
66
dcba
cbjid
+++
+=),(
cba
cbjid
++
+=),(
cba
ajisimJaccard
++
=),(
Name Gender Fever Cough Test-1 Test-2 Test-3 Test-4
Jack M Y N P N N N
Mary F Y N P N P N
Jim M Y P N N N N
75.0
211
21
),(
67.0
111
11
),(
33.0
102
10
),(
=
++
+
=
=
++
+
=
=
++
+
=
maryjimd
jimjackd
maryjackd
67.
One approach is to group each kind of variable together, performing a separate
cluster analysis for each variable type. This is feasible if these analyses derive
compatible results.
All variable types could be processed together for performing a single cluster
analysis. The different variables can be combined into a single dissimilarity
matrix, with a common scale of the interval [0.0, 1.0].
4.11 Categorization of Major Clustering Methods
The major categorization is:
1. Partitioning methods
2. Hierarchical methods
3. Density-based methods
4. Grid-based methods
5. Model-based clustering methods
6. Clustering high-dimensional data
7. Constraint-based cluster analysis
Partitioning Methods:
Given a database of n objects or data tuples, a partitioning method constructs k
partitions of the data, where each partition represents a cluster and k ≤ n.
It classifies the data into k groups, which together satisfy the following
requirements:
(1) Each group must contain at least one object and
(2) Each object must belong to exactly one group.
Given k, the number of partitions to construct, a partitioning method creates an
initial partitioning. It then uses an iterative relocation technique that attempts to
improve the partitioning by moving objects from one group to another. The
general criterion of a good partitioning is that objects in the same cluster are
“close” or related to each other, whereas objects of different clusters are “far apart”
or very different.
There are various kinds of other criteria for judging the quality of partitions.
Hierarchical Methods:
A hierarchical method creates a hierarchical decomposition of the given set of data
objects. A hierarchical method can be classified as being either agglomerative or
divisive, based on how the hierarchical decomposition is formed.
• The agglomerative approach- is also known as bottom-up approach. It starts
with each object forming a separate group. It successively merges the
objects or groups that are close to one another, until all of the groups are
merged into one or until a termination condition holds.
• The divisive approach – is known as top-down approach. It starts with all of
the objects in the same cluster and the cluster is split up into smaller
clusters, until eventually each object is in one cluster or until a termination
condition holds.
67
68.
Density-based Methods: A given cluster can be grown as long as the density in
the “neighborhood” exceeds some threshold; that is, for each data point within a
given cluster, the neighborhood of a given radius has to contain at least a minimum
number of points. Such a method can be used to filter out noise and discover
clusters of arbitrary shape.
Grid-based methods: Grid-based methods quantize the object space into a finite
number of cells that form a grid structure. Fast processing time, which is typically
independent of the number of data objects and dependent only on the number of
cells in each dimension in the quantized space is the merit of this method.
Model-based methods: Model-based methods hypothesize a model for each of the
clusters and find the best fit of the data to the given model. A model-based
algorithm may locate clusters by constructing a density function that reflects the
spatial distribution of the data points. It is used to automatically determine the
number of clusters based on standard statistics.
Clustering high-dimensional data: It is used for analysis of objects containing a
large number of features or dimensions. Frequent pattern–based clustering extracts
distinct frequent patterns among subsets of dimensions that occur frequently. It
uses such patterns to group objects and generate meaningful clusters.
Constraint-based clustering : It is a clustering approach that performs clustering
by incorporation of user-specified or application-oriented constraints. Various
kinds of constraints can be specified, either by a user or as per application
requirements.
4.12 Clustering Algorithms
Partitioning Methods : k-Means and k-Medoids
The most well-known and commonly used partitioning methods are k-means, k-
medoids, and their variations.
a. The k-Means Method
The k-means algorithm takes the input parameter, k, and partitions a set of n
objects into k clusters so that the resulting intra cluster similarity is high but the
inter cluster similarity is low. Cluster similarity is measured for the mean value of
the objects in a cluster, which can be viewed as the cluster’s centroid or center of
gravity.
• First, it randomly selects k of the objects, each of which initially represents
a cluster mean or center.
• For each of the remaining objects, an object is assigned to the cluster to
which it is the most similar, based on the distance between the object and
the cluster mean. It then computes the new mean for each cluster.
• This process iterates until the criterion function converges. Typically, the
square-error criterion is used, defined as
68
69.
Where E is the sum of the square error for all objects in the data set; p is the point
in space representing a given object; and mi is the mean of cluster Ci (both p and
mi are Multi dimensional). In other words, for each object in each cluster, the
distance from the object to its cluster center is squared, and the distances are
summed. This criterion tries to make the resulting k clusters as compact and as
separate as possible.
Fig. Clustering of a set of objects based on the k-means method.
b. Representative Object-Based Technique: The k-Medoids Method
The k-means algorithm is sensitive to outliers because an object with an extremely
large value may substantially distort the distribution of data.
Steps are:
• arbitrarily choose k objects in D as the initial representative objects or
seeds;
• repeat
o assign each remaining object to the cluster with the nearest
representative object;
o randomly select a non representative object
o Compute the total cost, S, of swapping representative object.
o if S < 0 then swap objects to form the new set of k representative
objects;
• until no change;
c. Partitioning Methods in Large Databases: From k-Medoids to CLARANS
CLARA (Clustering LARge Applications) can be used to deal with larger data
sets.
• Instead of taking the whole set of data into consideration, a small portion of
the actual data is chosen as a representative of the data.
• Medoids are then chosen from this sample using PAM.
• If the sample is selected in a fairly random manner, it should closely
represent the original data set.
• The representative objects (medoids) chosen will likely be similar to those
that would have been chosen from the whole data set.
69
70.
• CLARA draws multiple samples of the data set, applies PAM on each
sample, and returns its best clustering as the output.
The effectiveness of CLARA depends on the sample size. PAM searches for the
best k medoids among a given data set, whereas CLARA searches for the best k
medoids among the selected sample of the data set. CLARA cannot ﬁnd the best
clustering if any of the best sampled medoids is not among the best k medoids.
Hierarchical Methods
A hierarchical clustering method works by grouping data objects into a tree of
clusters.
a. Agglomerative and Divisive Hierarchical Clustering
Agglomerative hierarchical clustering: This bottom-up strategy starts by placing
each object in its own cluster and then merges these atomic clusters into larger and
larger clusters, until all of the objects are in a single cluster or until certain
termination conditions are satisfied.
Divisive hierarchical clustering: This top-down strategy does the reverse of
agglomerative hierarchical clustering by starting with all objects in one cluster. It
subdivides the cluster into smaller and smaller pieces, until each object forms a
cluster on its own or until it satisfies certain termination conditions, such as a
desired number of clusters is obtained or the diameter of each cluster is within a
certain threshold
b. BIRCH: Balanced Iterative Reducing and Clustering Using Hierarchies
BIRCH is designed for clustering a large amount of numerical data by integration
of hierarchical clustering and iterative partitioning.
The merits of BIRCH:
(1) scalability and
(2) the inability to undo what was done in the previous step.
BIRCH introduces two concepts, clustering feature and clustering feature tree (CF
tree), which are used to summarize cluster representations. These structures help
the clustering method achieve good speed and scalability in large databases and
also make it effective for incremental and dynamic clustering of incoming objects.
70
71.
Fig. 3.4.2.1 A CF tree structure.
C. ROCK: A Hierarchical Clustering Algorithm for Categorical Attributes
ROCK (RObust Clustering using linKs) is a hierarchical clustering algorithm that
explores the concept of links (the number of common neighbors between two
objects) for data with categorical attributes.
d. Chameleon: A Hierarchical Clustering Algorithm Using Dynamic
Modeling
Chameleon is a hierarchical clustering algorithm that uses dynamic modeling to
determine the similarity between pairs of clusters. It was derived based on the
observed weaknesses of two hierarchical clustering algorithms: ROCK and CURE.
ROCK and related schemes emphasize cluster interconnectivity while ignoring
information regarding cluster proximity. CURE and related schemes consider
cluster proximity yet ignore cluster interconnectivity.
In Chameleon, cluster similarity is assessed based on how well-connected objects
are within a cluster and on the proximity of clusters. That is, two clusters are
merged if their interconnectivity is high and they are close together. Thus,
Chameleon does not depend on a static, user-supplied model and can automatically
adapt to the internal characteristics of the clusters being merged. The merge
process facilitates the discovery of natural and homogeneous clusters and applies
to all types of data as long as a similarity function can be specified.
Fig. Chameleon: Hierarchical clustering based on k-nearest neighbors and dynamic
modeling.
Density-Based Methods
To discover clusters with arbitrary shape, density-based clustering methods have
been developed. These typically regard clusters as dense regions of objects in the
data space that are separated by regions of low density (representing noise).
DBSCAN grows clusters according to a density-based connectivity analysis.
OPTICS extends DBSCAN to produce a cluster ordering obtained from a wide
range of parameter settings. DENCLUE clusters objects based on a set of density
distribution functions.
a. DBSCAN
71
72.
DBSCAN (Density-Based Spatial Clustering of Applications with Noise) is a
density- based clustering algorithm. The algorithm grows regions with sufficiently
high density into clusters and discovers clusters of arbitrary shape in spatial
databases with noise. It defines a cluster as a maximal set of density-connected
points.
The basic ideas of density-based clustering involve a number of new deﬁnitions.
The steps involved are:
• The neighborhood within a radius ε of a given object is called the ε-
neighborhood of the object.
• If the ε-neighborhood of an object contains at least a minimum number,
MinPts, of objects, then the object is called a core object.
• Given a set of objects, D, we say that an object p is directly density-
reachable from object q if p is within the ε-neighborhood of q, and q is a
core object.
• An object p is density-reachable from object q with respect to ε and MinPts
in a set of objects, D, if there is a chain of objects p1, : : : , pn, where p1 = q
and pn = p such that pi+1 is directly density-reachable from pi.
• An object p is density-connected to object q with respect to ε and MinPts in
a set of objects, D, if there is an object o 2 D such that both p and q are
density-reachable from o with respect to ε and MinPts.
Fig. Density reachability and density connectivity in density-based clustering.
b. OPTICS: Ordering Points to Identify the Clustering Structure
OPTICS computes an augmented cluster ordering for automatic and interactive
cluster analysis. This ordering represents the density-based clustering structure of
the data. It contains information that is equivalent to density-based clustering
obtained from a wide range of parameter settings.
The cluster ordering can be used to extract basic clustering information (such as
cluster centers or arbitrary-shaped clusters) as well as provide the intrinsic
clustering structure.
72
73.
Fig. OPTICS terminology.
c. DENCLUE: Clustering Based on Density Distribution Functions
DENCLUE (DENsity-based CLUstEring) is a clustering method based on a set of
density distribution functions.
The method is built on the following ideas:
(1) The influence of each data point can be formally modeled using a mathematical
function, called an influence function, which describes the impact of a data point
within its neighborhood;
(2) The overall density of the data space can be modeled analytically as the sum of
the influence function applied to all data points; and
(3) Clusters can then be determined mathematically by identifying density
attractors, where density attractors are local maxima of the overall density
function.
Grid-Based Methods
The grid-based clustering approach uses a multi resolution grid data structure. It
quantizes the object space into a finite number of cells that form a grid structure on
which all of the operations for clustering are performed.
The main advantage of the approach is its fast processing time, which is typically
independent of the number of data objects, yet dependent on only the number of
cells in each dimension in the quantized space.
a. STING: STatistical INformation Grid
STING is a grid-based multi resolution clustering technique in which the spatial
area is divided into rectangular cells. There are usually several levels of such
rectangular cells corresponding to different levels of resolution, and these cells
form a hierarchical structure.
Each cell at a high level is partitioned to form a number of cells at the next lower
level. Statistical information regarding the attributes in each grid cell (such as the
mean, maximum, and minimum values) is pre computed and stored. These
statistical parameters are useful for query processing.
73
74.
Fig. A hierarchical structure for STING clustering.
b. Wave Cluster: Clustering Using Wavelet Transformation
Wave Cluster is a multi resolution clustering algorithm that summarizes the data
by imposing a multidimensional grid structure onto the data space. It then uses a
wavelet transformation to transform the original feature space, finding dense
regions in the transformed space.
In this approach, each grid cell summarizes the information of a group of points
that map into the cell. This summary information typically ﬁts into main memory
for use by the multi resolution wavelet transform and the subsequent cluster
analysis.
A wavelet transform is a signal processing technique that decomposes a signal into
different frequency sub bands. The wavelet model can be applied to d-dimensional
signals by applying a one-dimensional wavelet transforms d times. In applying a
wavelet transform, data are transformed so as to preserve the relative distance
between objects at different levels of resolution. This allows the natural clusters in
the data to become more distinguishable.
Model-Based Clustering Methods
Model-based clustering methods attempt to optimize the fit between the given data
and some mathematical model. Such methods are often based on the assumption
that the data are generated by a mixture of underlying probability distributions.
a. Expectation-Maximization
Expectation-Maximization is used to cluster the data using a finite mixture density
model of k probability distributions, where each distribution represents a cluster.
The problem is to estimate the parameters of the probability distributions so as to
best fit the data.
74
75.
Fig. Expectation- Maximization
Each cluster can be represented by a probability distribution, centered at a mean,
and with a standard deviation. Here, we have two clusters, corresponding to the
Gaussian distributions g(m1, σ1) and g(m2, σ2), respectively, where the dashed
circles represent the first standard deviation of the distributions.
The steps are:
1. Make an initial guess of the parameter vector
2. Iteratively refine the parameters (or clusters) based on the following two steps:
(a) Expectation Step: Assign each object xi to cluster Ck with the probability
(b) Maximization Step: Use the probability estimates from above to re-estimate (or
refine) the model parameters.
b. Conceptual Clustering
Conceptual clustering is a form of clustering in machine learning that, given a set
of unlabeled objects, produces a classification scheme over the objects. Unlike
conventional clustering, which primarily identifies groups of like objects,
conceptual clustering goes one step further by also finding characteristic
descriptions for each group, where each group represents a concept or class.
Hence, conceptual clustering process:
• Clustering is performed first, followed by characterization. Here, clustering
quality is not solely a function of the individual objects. Rather, it
incorporates factors such as the generality and simplicity of the derived
concept descriptions.
• Probabilistic descriptions are used to represent each derived concept.
•
COBWEB is a popular and simple method of incremental conceptual clustering.
Its input objects are described by categorical attribute-value pairs. COBWEB
creates a hierarchical clustering in the form of a classification tree.
c. Neural Network Approach
The neural network approach is motivated by biological neural networks.
75
76.
A neural network is a set of connected input/output units, where each connection
has a weight associated with it. Neural networks have several properties that make
them popular for clustering.
• Neural networks are inherently parallel and distributed processing
architectures.
• Neural networks learn by adjusting their interconnection weights so as to
best fit the data. This allows them to “normalize” or “prototype” the
patterns and act as feature (or attribute) extractors for the various clusters.
• Third, neural networks process numerical vectors and require object
patterns to be represented by quantitative features only.
• Many clustering tasks handle only numerical data or can transform their
data into quantitative features if needed.
The neural network approach to clustering tends to represent each cluster as an
exemplar. An exemplar acts as a “prototype” of the cluster and does not
necessarily have to correspond to a particular data example or object. New objects
can be distributed to the cluster whose exemplar is the most similar, based on some
distance measure.
The attributes of an object assigned to a cluster can be predicted from the attributes
of the Cluster’s exemplar. Self-organizing feature maps (SOMs) are one of the
most popular neural network methods for cluster analysis. They are sometimes
referred to as Kohonen self-organizing feature maps, after their creator, Teuvo
Kohonon, or as topologically ordered maps.
SOMs’ goal is to represent all points in a high-dimensional source space by points
in a low-dimensional (usually 2-D or 3-D) target space, such that the distance and
proximity relationships (hence the topology) are preserved as much as possible.
Clustering High – Dimensional Data
Most clustering methods are designed for clustering low-dimensional data and
encounter challenges when the dimensionality of the data grows really high. When
the dimensionality increases, usually only a small number of dimensions are
relevant to certain clusters, but data in the irrelevant dimensions may produce
much noise and mask the real clusters to be discovered.
When dimensionality increases, data usually become increasingly sparse because
the data points are likely located in different dimensional subspaces. When the
data become really sparse, data points located at different dimensions can be
considered as all equally distanced, and the distance measure, which is essential
for cluster analysis, becomes meaningless
a. CLIQUE: A Dimension-Growth Subspace Clustering Method
CLIQUE (CLustering InQUEst) was the first algorithm proposed for dimension-
growth subspace clustering in high-dimensional space. In dimension-growth
subspace clustering, the clustering process starts at single-dimensional subspaces
and grows upward to higher-dimensional ones. Because CLIQUE partitions each
76
77.
dimension like a grid structure and determines whether a cell is dense based on the
number of points it contains, it can also be viewed as an integration of density-
based and grid-based clustering methods. However, its overall approach is typical
of subspace clustering for high-dimensional space.
The ideas of the CLIQUE clustering algorithm are outlined as follows.
• Given a large set of multidimensional data points, the data space is usually
not uniformly occupied by the data points. CLIQUE’s clustering identifies
the sparse and the “crowded” areas in space, thereby discovering the overall
distribution patterns of the data set.
• A unit is dense if the fraction of total data points contained in it exceeds an
input model parameter. In CLIQUE, a cluster is defined as a maximal set of
connected dense units.
b. PROCLUS: A Dimension-Reduction Subspace Clustering Method
PROCLUS (PROjected CLUStering) is a dimension-reduction subspace clustering
method. It starts by finding an initial approximation of the clusters in the high-
dimensional attribute space. Each dimension is then assigned a weight for each
cluster, and the updated weights are used in the next iteration to regenerate the
clusters. This leads to the exploration of dense regions in all subspaces of some
desired dimensionality and avoids the generation of a large number of overlapped
clusters in projected dimensions of lower dimensionality.
PROCLUS finds the best set of medoids by a hill-climbing process similar to that
used in CLARANS, but generalized to deal with projected clustering. It adopts a
distance measure called Manhattan segmental distance, which is the Manhattan
distance on a set of relevant dimensions.
The PROCLUS algorithm consists of three phases:
• Initialization
• Iteration and
• Cluster refinement.
In the Initialization Phase, it uses a greedy algorithm to select a set of initial
medoids that are far apart from each other so as to ensure that each cluster is
represented by at least one object in the selected set.
It chooses a random sample of data points proportional to the number of clusters
to generate, and then applies the greedy algorithm to obtain an even smaller final
subset for the next phase.
The Iteration Phase selects a random set of k medoids from this reduced set (of
medoids), and replaces “bad” medoids with randomly chosen new medoids if the
clustering is improved. For each medoid, a set of dimensions is chosen whose
average distances are small compared to statistical expectation. The total number
of dimensions associated to medoids must be k.
77
78.
The Refinement Phase computes new dimensions for each medoid based on the
clusters found, reassigns points to medoids, and removes outliers.
c. Frequent Pattern–Based Clustering Methods
Frequent pattern mining searches for patterns (such as sets of items or objects) that
occur frequently in large data sets. Frequent pattern mining can lead to the
discovery of interesting associations and correlations among data objects.
Constraint-Based Cluster Analysis
Constraint-based clustering finds clusters that satisfy user-specified preferences or
constraints. Depending on the nature of the constraints, constraint-based clustering
may adopt rather different approaches.
Here are a few categories of constraints.
1. Constraints on individual objects:
This constraint confines the set of objects to be clustered. It can easily be handled
by preprocessing (e.g., performing selection using an SQL query), after which the
problem reduces to an instance of unconstrained clustering.
2. Constraints on the selection of clustering parameters:
A user may like to set a desired range for each clustering parameter. Clustering
parameters are usually quite specific to the given clustering algorithm. Examples
of parameters include k, the desired number of clusters in a k-means algorithm.
3. Constraints on distance or similarity functions:
Different distance or similarity functions can be specified for specific attributes of
the objects to be clustered or different distance measures for specific pairs of
objects.
a. Clustering with Obstacle Objects
During clustering, we must take obstacle objects into consideration. Obstacles
introduce constraints on the distance function. The straight-line distance between
two points is meaningless if there is an obstacle in the way.
b. Semi-Supervised Cluster Analysis
In comparison with supervised learning, clustering lacks guidance from users or
classifiers (such as class label information), and thus may not generate highly
desirable clusters. The quality of unsupervised clustering can be significantly
improved using some weak form of supervision, for example, in the form of pair
wise constraints.
A clustering process based on user feedback or guidance constraints is called semi-
supervised clustering.
Methods for semi-supervised clustering can be categorized into two classes:
• Constraint-Based Semi-Supervised Clustering and
• Distance-Based Semi-Supervised Clustering.
78
79.
Constraint-based semi-supervised clustering relies on user-provided labels or
constraints to guide the algorithm toward a more appropriate data partitioning.
This includes modifying the objective function based on constraints, or initializing
and constraining the clustering process based on the labeled objects.
Distance-based semi-supervised clustering employs an adaptive distance measure
that is trained to satisfy the labels or constraints in the supervised data.
4.13 Outlier Analysis
Data objects that do not comply with the general behavior or model of the data are
known as outliers.. Such data objects, which are grossly different from or
inconsistent with the remaining set of data, are called outliers.
Outliers can be caused by measurement or execution error. For example, the
display of a person’s age as 999 could be caused by a program default setting of
an unrecorded age. Outliers may be the result of inherent data variability.
Many data mining algorithms try to minimize the influence of outliers or eliminate
the mall together. This, however, could result in the loss of important hidden
information because one person’s noise could be another person’s signal.
It is useful for fraud detection, where outliers may indicate fraudulent activity.
Thus, outlier detection and analysis is an interesting data mining task, referred to
as outlier mining.
Outlier mining has wide applications. It can be used in fraud detection by detecting
unusual usage of credit cards or telecommunication services. In addition, it is
useful in customized marketing for identifying the spending behavior of customers
with extremely low or extremely high incomes, or in medical analysis for finding
unusual responses to various medical treatments.
The outlier mining problem can be viewed as two sub problems:
(1) Define what data can be considered as inconsistent in a given data set, and
(2) Find an efficient method to mine the outliers so defined.
3.5.1 Statistical Distribution-Based Outlier Detection
The statistical distribution-based approach to outlier detection assumes a
distribution or probability model for the given data set (e.g., a normal or Poisson
distribution) and then identifies outliers with respect to the model using a
discordancy test.
Application of the test requires knowledge of the data set parameters (such as the
assumed data distribution), knowledge of distribution parameters (such as the
mean and variance), and the expected number of outliers.
There are two basic types of procedures for detecting outliers:
• Block procedures: Either all of the suspect objects are treated as outliers or
all of them are accepted as consistent.
79
80.
• Consecutive (or sequential) procedures: An example of such a procedure is
the inside out procedure. The object that is least “likely” to be an outlier is
tested first. If it is found to be an outlier, then all of the more extreme values
are also considered outliers; otherwise, the next most extreme object is
tested, and so on. This procedure tends to be more effective than block
procedures.
3.5.2 Distance-Based Outlier Detection
The notion of distance-based outliers was used to overcome limitations imposed
by statistical methods. An object, o, in a data set, D, is a distance-based (DB)
outlier with parameters pct and dmin, that is, distance-based outliers as those
objects that do not have “enough” neighbors, where neighbors are defined based
on distance from the given object.
In comparison with statistical-based methods, distance based outlier detection
generalizes the ideas behind discordancy testing for various standard distributions.
Distance-based outlier detection avoids the excessive computation that can be
associated with fitting the observed distribution into some standard distribution
and in selecting discordancy tests.
Several efficient algorithms for mining distance-based outliers have been
developed.
These are outlined as follows
• Index-based algorithm: Given a data set, the index-based algorithm uses
multidimensional indexing structures, such as R-trees or k-d trees, to search
for neighbors of each object o within radius dmin around that object.
• Let M be the maximum number of objects within the dmin-neighborhood of
an outlier. Therefore, onceM+1 neighbors of object o are found, it is clear
that o is not an outlier. The index-based algorithm scales well as k
increases.
• Nested-loop algorithm: The nested-loop algorithm has the same
computational complexity as the index-based algorithm but avoids index
structure construction and tries to minimize the number of I/Os. It divides
the memory buffer space into two halves and the data set into several
logical blocks. By carefully choosing the order in which blocks are loaded
into each half, I/O efficiency can be achieved.
• Cell-based algorithm: To avoid O(n ) computational complexity, a cell-
based algorithm was developed for memory-resident data sets. Its
complexity is O(ck+n), where c is a constant depending on the number of
cells and k is the dimensionality
3.5.3 Density-Based Local Outlier Detection
Statistical and distance-based outlier detection both depend on the overall or
“global” distribution of the given set of data points, D.
These methods encounter difficulties when analyzing data with rather different
density distributions.
80
81.
To define the local outlier factor of an object, we need to introduce the concepts of
k-distance, k-distance neighborhood, reachability distance, and local reachability
density.
These are defined as follows:
The k-distance of an object p is the maximal distance that p gets from its k-nearest
neighbors. This distance is denoted as k-distance(p). It is defined as the distance,
d(p, o), between p and an object o ε D, such that
(1) For at least k objects, That is, there are at least k objects in D that are as close
as or closer to p than o, and
(2) There are at most k-1 objects, that are closer to p than o.
3.5.4 Deviation-Based Outlier Detection
Deviation-based outlier detection does not use statistical tests or distance-based
measures to identify exceptional objects. It identifies outliers by examining the
main characteristics of objects in a group. Objects that “deviate” from this
description are considered outliers.
a. Sequential Exception Technique
The sequential exception technique simulates the way in which humans can
distinguish unusual objects from among a series of supposedly like objects. It uses
implicit redundancy of the data. Given a data set, D, of n objects, it builds a
sequence of subsets,
fD1, D2, : : : , Dmg, of these objects with 2 µ m µ n such that
b. OLAP Data Cube Technique
An OLAP approach to deviation detection uses data cubes to identify regions of
anomalies in large multidimensional data. For added efficiency, the deviation
detection process is overlapped with cube computation.
The approach is a form of discovery-driven exploration, in which pre computed
measures indicating data exceptions are used to guide the user in data analysis, at
all levels of aggregation.
A cell value in the cube is considered an exception if it is significantly different
from the expected value, based on a statistical model. The method uses visual cues
such as background color to reflect the degree of exception of each cell.
The user can choose to drill down on cells that are flagged as exceptions. The
measure value of a cell may reflect exceptions occurring at more detailed or lower
levels of the cube, where these exceptions are not visible from the current level.
81
82.
UNIT V RECENT TRENDS
Multidimensional analysis and descriptive mining of complex data objects − Spatial
databases −Multimedia databases − Time series and sequence data − Text databases −
World Wide Web −Applications and trends in data mining.
5.1 MINING COMPLEX TYPES OF DATA
Introduction
Our previous studies on data mining techniques have focused on mining
relational data-bases, transactional databases, and data warehouses formed by the
transformation and integration of structured data. Vast amount of data in various
complex forms (e.g., structured and unstructured, hypertext and multimedia) have
been growing explosively owing to the rapid progress of data collection tolls,
advanced database system technologies and World –Wide Web (WWW)
82
83.
technologies. Therefore, an increasingly important task in data mining is to mine
complex types of data, including complex objects, spatial data, multimedia data,
time-series data, text data, and the World Wide Web.
In this chapter, we examine how to further develop the essential data
mining techniques (such as characterization, association, classification and
clustering), and how to develop new ones to cope with complex types of data and
perform fruitful knowledge mining in complex information repositories. Since
search into mining such complex databases has been evolving at a hasty pace, our
discussion covers only some preliminary issues. We expect that many books
dedicated to the mining of complex kinds of data will become available in the
future.
Multidimensional Analysis and Descriptive Mining of Complex Data Objects
A major limitation of many commercial data warehouse and OLAP tools
for multidimensional database analysis is their restriction on the allowable data
types for dimensions and measures. Most data cube implementations confine
dimensions to nonnumeric data and measures to simple aggregated values.
To introduce data mining and multidimensional data analysis for complex
objects, this section examines how to perform generalization on complex
structured objects and construct object cubes for OLAP and mining in object
databases.
The storage and access of complex structured data have been studied in object-
relational and object-oriented database systems. These systems organize a large set
of complex data objects into classes, which are in turn organized into
class/subclass hierarchies. Each object in a class is associated with
1) An object-identifier
2) A set of attributes that may contain sophisticated data structures, set- or list-
valued data, class composition and hierarchies, multimedia data and so on &
3) A set of methods that specify the computational routines or rules associated
with the object class.
To facilitate generalization and induction in object-relational and object-
oriented databases, it is important to study how the generalized data can be used
for multidimensional data and analysis and data mining.
5.2 Spatial Data Mining
Spatial data mining refers to the extraction of knowledge, spatial relationships, or
other interesting patterns not explicitly stored in spatial databases. Such mining
demands an integration of data mining with spatial database technologies.
A spatial database stores a large amount of space-related data, such as maps,
preprocessed remote sensing or medical imaging data, and VLSI chip layout data.
83
84.
Spatial databases have many features distinguishing them from relational
databases. They carry topological and/or distance information, usually organized
by sophisticated, multidimensional spatial indexing structures that are accessed by
spatial data access methods and often require spatial reasoning, geometric
computation, and spatial knowledge representation techniques.
It can be used for understanding spatial data, discovering spatial relationships and
relationships between spatial and non spatial data, constructing spatial knowledge
bases, reorganizing spatial databases, and optimizing spatial queries.
It is expected to have wide applications in geographic information systems,
geomarketing, remote sensing, image database exploration, medical imaging,
navigation, traffic control, environmental studies, and many other areas where
spatial data are used.
5.2.1 Spatial Data Cube Construction and Spatial OLAP
A spatial data warehouse is a subject-oriented, integrated, time-variant, and
nonvolatile collection of both spatial and non spatial data in support of spatial data
mining and spatial-data- related decision-making processes
5.2.2 Mining Spatial Association and Co-location Patterns
A spatial association rule is of the form AB [s%;c%], where A and B are sets of
spatial or non spatial predicates, s% is the support of the rule, and c% is the
confidence of the rule.
For example
This rule states that 80% of schools that are close to sports centers are also close to
parks, and 0.5% of the data belongs to such a case.
Examples include distance information, topological relations (like intersect,
overlap, and disjoint), and spatial orientations (like left of and west of).
Progressive refinement can be adopted in spatial association analysis.
• The method first mines large data sets roughly using a fast algorithm.
• It improves the quality of mining in a pruned data set using a more
expensive
algorithm.
• Superset coverage property is used to ensure that the pruned data set covers
the complete set of answers when applying the high-quality data mining
algorithms.
• It preserves all of the potential answers. In other words, it should allow a
false-positive test, which might include some data sets that do not belong to
the answer sets, but it should not allow a false-negative test, which might
exclude some potential answers.
84
85.
For mining spatial associations related to the spatial predicate close to, we can first
collect the candidates that pass the minimum support threshold by
1. Spatial Clustering Methods
Spatial data clustering identifies clusters, or densely populated regions, according
to some distance measurement in a large, multidimensional data set.
2. Spatial Classification and Spatial Trend Analysis
Spatial classification analyzes spatial objects to derive classification schemes in
relevance to certain spatial properties, such as the neighborhood of a district,
highway or river.
3. Mining Raster Databases
Spatial database systems usually handle vector data that consist of points, lines,
polygons (regions), and their compositions, such as networks or partitions. Typical
examples of such data include maps, design graphs, and 3-D representations of the
arrangement of the chains of protein molecules.
5.3 Multimedia Data Mining
A multimedia database system stores and manages a large collection of multimedia
data, such as audio, video, image, graphics, speech, text, document, and hypertext
data, which contain text, text markups, and linkages.
Multimedia database systems are increasingly common owing to the popular use
of audio-video equipment, digital cameras, CD-ROMs, and the Internet. Typical
multimedia database systems include NASA’s EOS (Earth Observation System),
various kinds of image and audio-video databases, and Internet databases.
5.3.1 Similarity Search in Multimedia Data
For similarity searching in multimedia data, we consider two main families of
multimedia indexing and retrieval systems:
(1) Description-Based Retrieval Systems, which build indices and perform object
retrieval based on image descriptions, such as keywords, captions, size, and time
of creation; and
(2) Content-Based Retrieval Systems, which support retrieval based on the image
content, such as color histogram, texture, pattern, image topology, and the shape of
objects and their layouts and locations within the image.
Description-based retrieval is labor-intensive if performed manually. If automated,
the results are typically of poor quality. For example, the assignment of keywords
to images can be a tricky and arbitrary task.
Content-based retrieval uses visual features to index images and promotes object
retrieval based on feature similarity, which is highly desirable in many
applications.
85
86.
In a content-based image retrieval system, there are often two kinds of queries:
• Image-sample-based queries and
• Image feature specification queries.
Image-sample-based queries
It is used to find all of the images that are similar to the given image sample. This
search compares the feature vector (or signature) extracted from the sample with
the feature vectors of images that have already been extracted and indexed in the
image database. Based on this comparison, images that are close to the sample
image are returned.
Image feature specification queries
It is used to specify or sketch image features like color, texture, or shape, which
are translated into a feature vector to be matched with the feature vectors of the
images in the database.
Content-based retrieval is used for medical diagnosis, weather prediction, TV
production, Web search engines for images, and e-commerce.
Similarity-based retrieval in image databases, based on image signature:
• Color histogram–based signature: In this approach, the signature of an
image includes color histograms based on the color composition of an
image regardless of its scale or orientation. This method does not contain
any information about shape, image topology, or texture. Thus, two images
with similar color composition but having very different shapes or textures
may be identified as similar, although they could be completely unrelated
semantically.
• Multi feature composed signature: In this approach, the signature of an
image includes a composition of multiple features: color histogram, shape,
image topology, and texture. The extracted image features are stored as
metadata, and images are indexed based on such metadata.
Multidimensional content-based search using Image features. It can therefore
be used to search for similar images.
• Wavelet-based signature: This approach uses the dominant wavelet
coefficients of an image as its signature. Wavelets capture shape, texture,
and image topology information in a single unified framework. This
improves efficiency and reduces the need for providing multiple search
primitives.
• Wavelet-based signature with region-based granularity: In this
approach, the computation and comparison of signatures are at the
granularity of regions, not the entire image. This is based on the observation
that similar images may contain similar regions, but a region in one image
86
87.
could be a translation or scaling of a matching region in the other.
Therefore, a similarity measure between the query image Q and a target
image T can be defined in terms of the fraction of the area of the two
images covered by matching pairs of regions from Q and T. Such a region-
based similarity search can find images containing similar objects, where
these objects may be translated or scaled.
5.3.2 Multidimensional Analysis of Multimedia Data
A multimedia data cube can contain additional dimensions and measures for
multimedia information, such as color, texture, and shape.
5.3.3 Classification and Prediction Analysis of Multimedia Data
Classification and predictive modeling have been used forming multimedia data,
especially in scientific research, such as astronomy, seismology.
5.3.4 Mining Associations in Multimedia Data
Association rules involving multimedia objects can be mined in image and video
databases.
Three categories can be observed:
• Associations between image content and non image content features: A rule
like “If at least 50% of the upper part of the picture is blue, then it is likely
to represent sky” belongs to this category since it links the image content to
the keyword sky.
• Associations among image contents that are not related to spatial
relationships: A rule like “If a picture contains two blue squares, then it is
likely to contain one red circle as well” belongs to this category since the
associations are all regarding image contents.
• Associations among image contents related to spatial relationships: A rule
like “If a red triangle is between two yellow squares, then it is likely a big
oval-shaped object is underneath” belongs to this category since it
associates objects in the image with spatial relationships.
To mine associations among multimedia objects, we can treat each image as a
transaction and find frequently occurring patterns among different images.
5.3.5 Audio and Video Data Mining
Besides still images, an incommensurable amount of audio visual information is
becoming available in digital form, in digital archives, on the World Wide Web, in
broadcast data streams, and in personal and professional databases.
Typical examples include searching for and multimedia editing of particular video
clips in a TV studio, detecting suspicious persons or scenes in surveillance videos,
searching for particular events in a personal multimedia repository such as
87
88.
MyLifeBits, discovering patterns and outliers in weather radar recordings, and
finding a particular melody or tune in MP3 audio album.
5.4 Mining Time-Series Data
Time-series database
Consists of sequences of values or events changing with time
Data is recorded at regular intervals
Characteristic time-series components
Trend, cycle, seasonal, irregular
Applications
Financial: stock price, inflation
Industry: power consumption
Scientific: experiment results
Meteorological: precipitation
Categories of Time-Series Movements
Long-term or trend movements (trend curve): general direction in
which a time series is moving over a long interval of time
Cyclic movements or cycle variations: long term oscillations about a
trend line or curve
88
89.
e.g., business cycles, may or may not be periodic
Seasonal movements or seasonal variations
i.e, almost identical patterns that a time series appears to
follow during corresponding months of successive years.
Irregular or random movements
Time series analysis: decomposition of a time series into these four basic
movements
Additive Modal: TS = T + C + S + I
Multiplicative Modal: TS = T × C × S × I
The freehand method
Fit the curve by looking at the graph
Costly and barely reliable for large-scaled data mining
The least-square method
Find the curve minimizing the sum of the squares of the deviation of
points on the curve from the corresponding data points
The moving-average method
Moving average of order n
Smoothes the data
Eliminates cyclic, seasonal and irregular movements
Loses the data at the beginning or end of a series
Sensitive to outliers (can be reduced by weighted moving average)
Seasonal index
Set of numbers showing the relative values of a variable during the
months of the year
E.g., if the sales during October, November, and December are 80%,
120%, and 140% of the average monthly sales for the whole year,
respectively, then 80, 120, and 140 are seasonal index numbers for
these months
Deseasonalized data
Data adjusted for seasonal variations for better trend and cyclic
analysis
Divide the original monthly data by the seasonal index numbers for
the corresponding months
Estimation of cyclic variations
If (approximate) periodicity of cycles occurs, cyclic index can be
constructed in much the same manner as seasonal indexes
Estimation of irregular variations
By adjusting the data for trend, seasonal and cyclic variations
With the systematic analysis of the trend, cyclic, seasonal, and irregular
components, it is possible to make long- or short-term predictions with
reasonable quality
89
90.
5.5 Text Databases and IR
Text databases (document databases)
Large collections of documents from various sources: news articles,
research papers, books, digital libraries, e-mail messages, and Web
pages, library database, etc.
Data stored is usually semi-structured
Traditional information retrieval techniques become inadequate for
the increasingly vast amounts of text data
Information retrieval
A field developed in parallel with database systems
Information is organized into (a large number of) documents
Information retrieval problem: locating relevant documents based on
user input, such as keywords or example documents
Information Retrieval
Typical IR systems
Online library catalogs
Online document management systems
Information retrieval vs. database systems
Some DB problems are not present in IR, e.g., update, transaction
management, complex objects
Some IR problems are not addressed well in DBMS, e.g.,
unstructured documents, approximate search using keywords and
relevance
90
91.
Keyword-Based Retrieval
A document is represented by a string, which can be identified by a set of
keywords
Queries may use expressions of keywords
E.g., car and repair shop, tea or coffee, DBMS but not Oracle
Queries and retrieval should consider synonyms, e.g., repair and
maintenance
Major difficulties of the model
Synonymy: A keyword T does not appear anywhere in the document,
even though the document is closely related to T, e.g., data mining
Polysemy: The same keyword may mean different things in different
contexts, e.g., mining
Similarity-Based Retrieval in Text Databases
Finds similar documents based on a set of common keywords
Answer should be based on the degree of relevance based on the nearness
of the keywords, relative frequency of the keywords, etc.
Basic techniques
Stop list
91
92.
Set of words that are deemed “irrelevant”, even though they
may appear frequently
E.g., a, the, of, for, with, etc.
Stop lists may vary when document set varies
5.6 Mining the World-Wide Web
The WWW is huge, widely distributed, global information service center
for
Information services: news, advertisements, consumer information,
financial management, education, government, e-commerce, etc.
Hyper-link information
Access and usage information
WWW provides rich sources for data mining
Challenges
Too huge for effective data warehousing and data mining
Too complex and heterogeneous: no standards and structure
Web search engines
Index-based: search the Web, index Web pages, and build and store huge
keyword-based indices
Help locate sets of Web pages containing certain keywords
92
93.
Deficiencies
A topic of any breadth may easily contain hundreds of thousands of
documents
Many documents that are highly relevant to a topic may not contain
keywords defining them (polysemy)
HITS (Hyperlink-Induced Topic Search)
Explore interactions between hubs and authoritative pages
Use an index-based search engine to form the root set
Many of these pages are presumably relevant to the search topic
Some of them should contain links to most of the prominent
authorities
Expand the root set into a base set
Include all of the pages that the root-set pages link to, and all of the
pages that link to a page in the root set, up to a designated size cutoff
Apply weight-propagation
An iterative process that determines numerical estimates of hub and
authority weights
XML can help to extract the correct descriptors
Standardization would greatly facilitate information extraction
Potential problem
93
94.
XML can help solve heterogeneity for vertical applications,
but the freedom to define tags can make horizontal
applications on the Web more heterogeneous
5.7 Data mining applications
Data mining is a young discipline with wide and diverse applications
There is still a nontrivial gap between general principles of data
mining and domain-specific, effective data mining tools for
particular applications
Some application domains (covered in this chapter)
Biomedical and DNA data analysis
Financial data analysis
Retail industry
Telecommunication industry
DNA sequences: 4 basic building blocks (nucleotides): adenine (A),
cytosine (C), guanine (G), and thymine (T).
Gene: a sequence of hundreds of individual nucleotides arranged in a
particular order
Humans have around 100,000 genes
Tremendous number of ways that the nucleotides can be ordered and
sequenced to form distinct genes
94
95.
Semantic integration of heterogeneous, distributed genome databases
Current: highly distributed, uncontrolled generation and use of a
wide variety of DNA data
Similarity search and comparison among DNA sequences
Compare the frequently occurring patterns of each class (e.g.,
diseased and healthy)
Identify gene sequence patterns that play roles in various diseases
Association analysis: identification of co-occurring gene sequences
Most diseases are not triggered by a single gene but by a
combination of genes acting together
Association analysis may help determine the kinds of genes that are
likely to co-occur together in target samples
Path analysis: linking genes to different disease development stages
Different genes may become active at different stages of the disease
Develop pharmaceutical interventions that target the different stages
separately
Visualization tools and genetic data analysis
Data cleaning and data integration methods developed in data mining
will help
Data Mining for Financial Data Analysis
Financial data collected in banks and financial institutions are often
relatively complete, reliable, and of high quality
Design and construction of data warehouses for multidimensional data
analysis and data mining
View the debt and revenue changes by month, by region, by sector,
and by other factors
Access statistical information such as max, min, total, average, trend,
etc.
Loan payment prediction/consumer credit policy analysis
feature selection and attribute relevance ranking
Loan payment performance
Consumer credit rating
Financial Data Mining
Classification and clustering of customers for targeted marketing
multidimensional segmentation by nearest-neighbor, classification,
decision trees, etc. to identify customer groups or associate a new
customer to an appropriate customer group
Detection of money laundering and other financial crimes
95
96.
integration of from multiple DBs (e.g., bank transactions,
federal/state crime history DBs)
Tools: data visualization, linkage analysis, classification, clustering tools, outlier
analysis, and sequential pattern analysis tools (find unusual access sequences
96
Views
Actions
Embeds 0
Report content