Transcript Lecture notes for chapter 2 (Powerpoint Slideshow file)

Data Mining and Data Warehousing
Introduction
 Data warehousing and OLAP for data mining
 Data preprocessing
 Primitives for data mining
 Concept description
 Mining association rules in large databases
 Classification and prediction
 Cluster analysis
 Mining complex types of data
 Applications and trends in data mining

Copyright Jiawei Han. Modified by
1
Data Mining and Warehousing: Session 2
Data Warehouse and OLAP
Technology for Data Mining
Copyright Jiawei Han. Modified by
2
Session 2: Data Warehousing and OLAP
Technology for Data Mining

What is a data warehouse?

A multi-dimensional data model

Data warehouse architecture

Data warehouse implementation

Further development of data cube technology

From data warehousing to data mining
Copyright Jiawei Han. Modified by
3
Motivation
Data has been collected everywhere and in huge
amounts — how to make good use of your data?
 Bring together scattered information from multiple
sources as to provide a consistent database source
for decision support.
 Provide tools for business executives to
systematically organize, understand, and use their
data to make strategic decisions.

Copyright Jiawei Han. Modified by
4
What is a Data Warehouse?

Defined in many different ways, but not rigorously.


A decision support database that is maintained
separately from the organization’s operational database
Support information processing by providing a solid
platform of consolidated, historical data for analysis.
“A data warehouse is a subject-oriented,
integrated, time-variant, and nonvolatile collection
of data in support of management’s decisionmaking process.” — W. H. Inmon
 Data warehousing:


The process of constructing and using data warehouses
Copyright Jiawei Han. Modified by
5
Data Warehouse — Subject Oriented

Organized around major subjects of interests, such
as profit, product, sales.

Focusing on the modeling and analysis of data for
decision makers, not on daily operations or
transaction processing.

Provide a simple and concise view around
particular subject issues by excluding data that are
not useful in the decision support process.
Copyright Jiawei Han. Modified by
6
Data Warehouse — Integrated

Integrate multiple, heterogeneous data sources


relational databases, flat files, on-line transaction records
Data cleaning and data integration techniques are
applied.

Ensure consistency in naming conventions, encoding
structures, attribute measures, etc. among different data
sources
– E.g., Hotel price: currency, tax, breakfast covered, etc.

When data is moved to the warehouse, it is converted.
Copyright Jiawei Han. Modified by
7
Data Warehouse — Time Variant


The time horizon for the data warehouse is
significantly longer than that of operational systems.

Operational database: current value data.

Data warehouse data: provide information from a
historical perspective (e.g., past 5-10 years)
Every key structure in the data warehouse

Contains an element of time, explicitly or implicitly
Copyright Jiawei Han. Modified by
8
Data Warehouse — Non-Volatile

A physically separate store of data transformed from
the operational environment.

Operational update of data does not occur in the
data warehouse environment. (May update monthly)

Does not require transaction processing, recovery, and
concurrency control mechanisms

Requires only two operations in data accessing:
– initial loading of data and access of data.
Copyright Jiawei Han. Modified by
9
Data Warehouse vs. Heterogeneous DBMS

Traditional heterogeneous DB integration:

Build wrappers/mediators on top of heterogeneous
databases

Query driven approach:
– When a query is posed to a client site, a meta-dictionary
is used to translate the query into queries appropriate
for individual heterogeneous sites involved, and the
results are integrated into a global answer set.
– Complex information filtering, compete for sources

Data warehouse: high performance, simpler to use
Copyright Jiawei Han. Modified by
10
Data Warehouse vs. Operational DB Systems

Major task of traditional relational DBMS



OLTP (on-line transaction processing)
Day-to-day operations: purchasing, inventory, banking,
manufacturing, payroll, registration, accounting, etc.
Major task of data warehouse system


OLAP (on-line analytical processing)
Data analysis and decision making
Copyright Jiawei Han. Modified by
11
OLTP vs. OLAP
OLTP
OLAP
users
Clerk, IT professional
Knowledge worker
function
day to day operations
decision support
DB design
application-oriented
subject-oriented
data
current, up-to-date
detailed, flat relational
isolated
repetitive
historical,
summarized, multidimensional
integrated, consolidated
ad-hoc
lots of scans
unit of work
read/write
index/hash on prim. key
short, simple transaction
# records accessed
tens
millions
#users
thousands
hundreds
DB size
100MB-GB
100GB-TB
metric
transaction throughput
query throughput, response
usage
access
complex query
Copyright Jiawei Han. Modified by
12
Why Separate Data Warehouse?

High performance for both systems:



DBMS — tuned for OLTP: access methods, indexing,
concurrency control, recovery
Warehouse — tuned for OLAP: complex OLAP queries,
multidimensional view, consolidation.
Different functions and different data:



data source: Decision support requires historical data
which operational DBs do not typically maintain
data consolidation: DS requires consolidation
(aggregation, summarization) of data from heterogeneous
sources
data quality: different sources typically use inconsistent
data representations, codes and formats which have to be
reconciled.
Copyright Jiawei Han. Modified by
13
Session 2: Data Warehousing and OLAP

What is a data warehouse?

A multi-dimensional data model

Data warehouse architecture

Data warehouse implementation

Further development of data cube technology

From data warehousing to data mining
Copyright Jiawei Han. Modified by
14
Multidimensional Data for OLAP
Sales volume as a function of product, month, and
region
Product

Month
Copyright Jiawei Han. Modified by
15
Aggregate with lower dimension Cuboids
all
0-D(apex) cuboid
product
product,date
date
country
product,country
1-D cuboids
date, country
2-D cuboids
3-D(base) cuboid
product, date, country
Copyright Jiawei Han. Modified by
16
Cube: A Lattice of Cuboids
all
time
time,item
0-D(apex) cuboid
item
time,location
location
item,location
time,supplier
time,item,location
supplier
1-D cuboids
location,supplier
2-D cuboids
item,supplier
time,location,supplier
3-D cuboids
time,item,supplier
item,location,supplier
4-D(base) cuboid
time, item, location, supplier
Copyright Jiawei Han. Modified by
17
A concept hierarchy:
Aggregate on dimension (location)
all
all
Europe
region
country
city
office
Germany
Frankfurt
...
...
...
Spain
North_America
Canada
Vancouver ...
L. Chan
...
Copyright Jiawei Han. Modified by
...
Mexico
Toronto
M. Young
18
A Sample Data Cube
2Qtr
3Qtr
4Qtr
sum
U.S.A
Canada
Mexico
Country
TV
PC
VCR
sum
1Qtr
Date
Total annual sales
of TV in U.S.A.
sum
Copyright Jiawei Han. Modified by
19
Measures: Three Categories of Measures

distributive: if the result derived by applying the function to n
aggregate values is the same as that derived by applying the
function on all the data without partitioning.
– E.g., count(), sum(), min(), max().

algebraic: if it can be computed by an algebraic function with
M arguments (where M is a bounded integer), each of which is
obtained by applying a distributive aggregate function.
– E.g., avg(), min_N(), standard_deviation().

holistic: if there is no constant bound on the storage size
needed to describe a subaggregate.
– E.g., median(), mode(), rank().
Copyright Jiawei Han. Modified by
20
Typical OLAP Operations





Roll up (drill-up): summarize data
 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 to its back-end
relational tables.Copyright Jiawei Han. Modified by
21
Conceptual Modeling of Data Warehouses

Modeling data warehouses: dimensions &
measurements

Star schema: A single object (fact table) in the middle
connected to a number of objects (dimension tables)

Snowflake schema: A refinement of star schema where the
dimensional hierarchy is represented explicitly by
normalizing the dimension tables.

Fact constellations: Multiple fact tables share dimension
tables.
Copyright Jiawei Han. Modified by
22
Example of Star Schema
Date
Product
Day
Month
Year
ProductNo
ProdName
ProdDesc
Category
QOH
Sales Fact Table
Date
Product
Store
StoreID
City
State
Country
Region
Store
Cust
Customer
unit_sales
dollar_sales
CustId
CustName
CustCity
CustCountry
Yen_sales
Measurements
Copyright Jiawei Han. Modified by
23
Example of Snowflake Schema
Year
Year
Product
Month
Month
Year
Date
Sales Fact Table
Day
Month
Date
Product
Store
City
City
State
State
Country
Country
Region
StoreID
City
Store
Cust
Customer
unit_sales
dollar_sales
State
Country
ProductNo
ProdName
ProdDesc
Category
QOH
CustId
CustName
CustCity
CustCountry
Yen_sales
Measurements
Copyright Jiawei Han. Modified by
24
A Star-Net Query Model
Customer Orders
Shipping Method
Customer
CONTRACTS
AIR-EXPRESS
ORDER
TRUCK
PRODUCT LINE
Time
Product
ANNUALY QTRLY
DAILY
PRODUCT ITEM PRODUCT GROUP
CITY
SALES PERSON
COUNTRY
DISTRICT
REGION
DIVISION
Location
Promotion
Copyright Jiawei Han. Modified by
Organization
25
Session 2: Data Warehousing and OLAP
Technology for Data Mining

What is a data warehouse?

A multi-dimensional data model

Data warehouse architecture

Data warehouse implementation

Further development of data cube technology

From data warehousing to data mining
Copyright Jiawei Han. Modified by
26
Design of a Data Warehouse: A Business
Analysis Framework





Four different views regarding the design of a data
warehouse must be considered
Top-down view:
 Allows selection of the relevant information necessary
for the data warehouse.
Data source view:
 exposes the information being captured, stored, and
managed by operational systems
Data warehouse view:
 fact tables and dimension tables
Business query view:
 perspectives of data in the warehouse from the view of
end-user.
Copyright Jiawei Han. Modified by
27
The Process of Data Warehouse Design

Top-down, bottom-up approaches or a combination of both



From software engineering point of view



Top-down: Starts with overall design and planning (mature)
Bottom-up: Starts with experiments and prototypes (rapid)
Waterfall: structured and systematic analysis at each step before
proceeding to the next.
Spiral: rapid generation of increasingly functional systems, short
turn around time, quick turn around.
Typical data warehouse design process:




Choose a business process to model, e.g., orders, invoices, etc.
Choose the grain (atomic level of data) of the business process
Choose the dimensions that will apply to each fact table record
Choose the measure that will populate each fact table record.
Copyright Jiawei Han. Modified by
28
Multi-Tiered Architecture
other
Metadata
sources
Operational
DBs
Extract
Transform
Load
Refresh
Monitor
&
Integrator
Data
Warehouse
OLAP Server
Serve
Analysis
Query
Reports
Data mining
Data Marts
Data Sources
Data Storage
OLAP Engine Front-End Tools
Copyright Jiawei Han. Modified by
29
Three-Tier Data Warehouse Architecture

Enterprise warehouse:


collects all of the information about subjects spanning the
entire organization.
Data Mart:

a subset of corporate-wide data that is of value to a specific
groups of users. Its scope is confined to specific, selected
groups, such as marketing data mart.
– Independent vs. dependent (directly from warehouse) data mart

Virtual warehouse:


A set of views over operational databases.
Only some of the possible summary views may be
materialized.
Copyright Jiawei Han. Modified by
30
Data Warehouse Development: A
Recommended Approach
Multi-Tier
Data
Warehouse
Distributed
Data Marts
Data
Mart
Data
Mart
Model refinement
Enterprise
Data
Warehouse
Model refinement
Define a high-level corporate data model
Copyright Jiawei Han. Modified by
31
OLAP Server Architectures

Relational OLAP (ROLAP):




Multidimensional OLAP (MOLAP):



Array-based multidimensional storage engine (sparse matrix
techniques)
fast indexing to pre-computed summarized data
Hybrid OLAP (HOLAP):


Use relational or extended-relational DBMS to store and manage
warehouse data and OLAP middle ware to support missing pieces.
Include optimization of DBMS backend, implementation of
aggregation navigation logic, and additional tools and services
greater scalability
User flexibility, e.g., low level: relational, high-level: array.
Specialized SQL servers:

specialized support for SQL queries over star.snowflake schemas
Copyright Jiawei Han. Modified by
32
Session 2: Data Warehousing and OLAP
Technology for Data Mining

What is a data warehouse?

A multi-dimensional data model

Data warehouse architecture

Data warehouse implementation

Further development of data cube technology

From data warehousing to data mining
Copyright Jiawei Han. Modified by
33
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), none, or some.
Algorithms for selection of which cuboids to materialize.
– Based on size, sharing, access frequency, etc.

Efficient cube computation methods


ROLAP-based cubing algorithms.
Array-based cubing algorithm.
Copyright Jiawei Han. Modified by
34
Cube Operation

Cube definition and computation in DMQL
define cube sales[item, city, year]: sum(sales_in_dollars)
computer 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
(city)
(item)
(year)
CUBE BY item, city, year

Need compute the following Group-Bys
(city, item)
(city, year)
(item, year)
(date, product, customer),
(date,product),(date, customer), (product, customer),
(date), (product), (customer)
(city, item, year)
()
Copyright Jiawei Han. Modified by
35
Cube Computation: ROLAP-Based Method

Hash/sort based methods (Agarwal et. al. VLDB’96)





Smallest-parent: computing a cuboid from the smallest
cubod previously computed cuboid.
Cache-results: caching results of a cuboid from which
other cuboids are computed to reduce disk I/Os.
Amortize-scans: computing as many as possible cuboids at
the same time to amortize disk reads.
Share-sorts: sharing sorting costs cross multiple cuboids
when sort-based method is used.
Share-partitions: sharing the partitioning cost cross
multiple cuboids when hash-based algorithms are used.
Copyright Jiawei Han. Modified by
36
Multi-way Array Aggregation for Cube
Computation
Partition arrays into chunks (small subcube fit in memory).
 Compressed sparse array addressing: (chunk_id, offset)
 Compute aggregates in “multiway” by visiting cube cells in
the order which minimize # of times to visit each cell, reduce
memory access and storage cost.
62
63
64
C c2 c3 61
45
46
47
48
c1 29
30
31
32
What is the best
c0
60
traversing order
14
15
16
b3 B13
44
to do multiway
28 56
b2 9
40
aggregation?
B
24

b1
5
b0
1
2
3
4
a0
a1
a2
a3
A
36
20
52
Copyright Jiawei Han. Modified by
37
Multi-way Array Aggregation for Cube
Computation
C
c3 61
62
63
64
c2 45
46
47
48
c1 29
30
31
32
c0
b3
B
b2
B13
14
15
60
16
44
28
9
24
b1
5
b0
1
2
3
4
a0
a1
a2
a3
56
40
36
52
20
A
Copyright Jiawei Han. Modified by
38
Multi-way Array Aggregation for Cube
Computation
C
c3 61
62
63
64
c2 45
46
47
48
c1 29
30
31
32
c0
b3
B
b2
B13
14
15
60
16
44
28
9
24
b1
5
b0
1
2
3
4
a0
a1
a2
a3
56
40
36
52
20
A
Copyright Jiawei Han. Modified by
39
Multi-way Array Aggregation for Cube
Computation(Cont.)

Conclusion: the planes should be sorted and
computed according to their size in ascending
order.


E.g. if |AB| < |AC| < |BC| then first compute AB then AC
then BC.
The same for 1-D planes.

E.g. if |A| < |B| < |C| then first compute A then B then C.
Copyright Jiawei Han. Modified by
40
Indexing OLAP Data: Bitmap Index
Index on a particular column
 Each value in the column has a bit vector: bit-op is fast
 The length of the bit vector: # of records in the base table
 The i-th bit is set if the i-th row of the base table has the
value for the indexed column
 not suitable for high cardinality domains
Base table
Index on Region
Index on Type

Cust
C1
C2
C3
C4
C5
Region
Asia
Europe
Asia
America
Europe
Type RecIDAsia Europe America RecID Retail Dealer
Retail
1
1
0
0
1
1
0
Dealer 2
0
1
0
2
0
1
Dealer 3
1
0
0
3
0
1
4
0
1
0
4
1
0
Retail
0
0
1
5
0
1
Dealer 5
Copyright Jiawei Han. Modified by
41
Indexing OLAP Data: Join Indexes

Traditional indexes map the vales to a list of record ids.

Join indexes map the tuples in the join result of two
relations to the source tables.

In data warehouse cases, join indexes relate the values of the
dimensions of a start schema to rows in the fact table.

For a warehouse with a Sales fact table and two
dimensions city and product, a join index on city
maintains for each distinct city a list of RIDs of the tuples
recording the sales in the city

Join indexes can span multiple dimensions
Copyright Jiawei Han. Modified by
42
Efficient Processing OLAP Queries

Determine which operations should be performed on the
available cuboids:

transform drill, roll, etc. infro corresponding SQL and/or
OLAP operations, e.g, dice = selection + projection

Determine to which materialized cuboid(s) the relevant
operations should be applied.

Exploring indexing structures and compressed vs. dense
array structures in MOLAP
Copyright Jiawei Han. Modified by
43
Metadata Repository


Meta data are the data defining warehouse objects
Description the structure of the warehouse


Operational meta-data




data lineage (history of migrated data and transformation path),
currency of data (active, archived, or purged), monitoring
information (warehouse usage statistics, error reports, audit trails)
The algorithms used for summarization
The mapping from operational environment to the data
warehouse
Data related to system performance


schema, view, dimensions, hierarchies, derived data defn, data mart
locations and contents
warehouse schema, view and derived data definitions
Business data

business terms and Copyright
definitions,
ownership
Jiawei Han.
Modified byof data, charging policies
44
Data Warehouse Back-End Tools and Utilities





Data extraction:
 get data from multiple, heterogeneous, and external
sources
Data cleaning:
 detect errors in the data and rectify them when possible
Data transformation:
 convert data from legacy or host format to warehouse
format
Load:
 sort, summarize, consolidate, compute views, check
integrity, and build indicies and partitions
Refresh
 propagate the updates from the data sources to the
warehouse
Copyright Jiawei Han. Modified by
45
Session 2: Data Warehousing and OLAP
Technology for Data Mining

What is a data warehouse?

A multi-dimensional data model

Data warehouse architecture

Data warehouse implementation

Further development of data cube technology

From data warehousing to data mining
Copyright Jiawei Han. Modified by
46
Discovery-Driven Exploration of Data Cubes

Hypothesis-driven: exploration by user, huge search space

Discovery-driven (Sarawagi et al.’98):

pre-compute measures indicating exceptions, guide user
in the data analysis, at all levels of aggregation.

Exception: significantly different from the value
anticipated, based on a statistical model.

Visual cues such as background color are used to reflect
the degree of exception of each cell.

Computation of exception indicator (modeling fitting and
computing SelfExp, InExp, and PathExp values) can be
overlapped with cube construction
Copyright Jiawei Han. Modified by
47
Session 2: Data Warehousing and OLAP
Technology for Data Mining

What is a data warehouse?

A multi-dimensional data model

Data warehouse architecture

Data warehouse implementation

Further development of data cube technology

From data warehousing to data mining
Copyright Jiawei Han. Modified by
48
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
Copyright Jiawei Han. Modified by
49
From On-Line Analytical Processing to On
Line Analytical Mining

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.

Architecture of OLAM
Copyright Jiawei Han. Modified by
50
An OLAM Architecture
Mining query
Mining result
Layer4
User Interface
User GUI API
OLAM
Engine
OLAP
Engine
Layer3
OLAP/OLAM
Data Cube API
Layer2
MDDB
MDDB
Meta Data
Filtering&Integration
Database API
Filtering
Layer1
Data cleaning
Databases
Data
Data integration Warehouse
Copyright Jiawei Han. Modified by
Data
Repository
51