This action might not be possible to undo. Are you sure you want to continue?

BooksAudiobooksComicsSheet Music### Categories

### Categories

### Categories

Editors' Picks Books

Hand-picked favorites from

our editors

our editors

Editors' Picks Audiobooks

Hand-picked favorites from

our editors

our editors

Editors' Picks Comics

Hand-picked favorites from

our editors

our editors

Editors' Picks Sheet Music

Hand-picked favorites from

our editors

our editors

Top Books

What's trending, bestsellers,

award-winners & more

award-winners & more

Top Audiobooks

What's trending, bestsellers,

award-winners & more

award-winners & more

Top Comics

What's trending, bestsellers,

award-winners & more

award-winners & more

Top Sheet Music

What's trending, bestsellers,

award-winners & more

award-winners & more

Welcome to Scribd! Start your free trial and access books, documents and more.Find out more

Data Mining:

Concepts and Techniques

April 30, 2012 2

Chapter 5: Concept Description:

Characterization and Comparison

What is concept description?

Data generalization and summarization-based

characterization

Analytical characterization: Analysis of attribute relevance

Mining class comparisons: Discriminating between

different classes

Mining descriptive statistical measures in large databases

Discussion

Summary

What is Concept 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

April 30, 2012 4

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

April 30, 2012 5

Chapter 5: Concept Description:

Characterization and Comparison

What is concept description?

Data generalization and summarization-based

characterization

Analytical characterization: Analysis of attribute relevance

Mining class comparisons: Discriminating between

different classes

Mining descriptive statistical measures in large databases

Discussion

Summary

April 30, 2012 6

Data 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

1

2

3

4

5

Conceptual levels

April 30, 2012 7

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

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

April 30, 2012 8

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. see example

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.

Presentation: User interaction: (1) adjust levels by drilling,

(2) pivoting, (3) mapping into rules, cross tabs,

visualization presentations.

See Implementation See example See complexity

April 30, 2012 11

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‖ }

Class Characterization: An Example

Name Gender Major Birth-Place Birth_date Residence Phone # GPA

Jim

Woodman

M CS Vancouver,BC,

Canada

8-12-76 3511 Main St.,

Richmond

687-4598 3.67

Scott

Lachance

M CS Montreal, Que,

Canada

28-7-75 345 1st Ave.,

Richmond

253-9106 3.70

Laura Lee

…

F

…

Physics

…

Seattle, WA, USA

…

25-8-70

…

125 Austin Ave.,

Burnaby

…

420-5232

…

3.83

…

Removed Retained Sci,Eng,

Bus

Country Age range City Removed Excl,

VG,..

Gender Major Birth_region Age_range Residence GPA Count

M Science Canada 20-25 Richmond Very-good 16

F Science Foreign 25-30 Burnaby Excellent 22

… … … … … … …

Birth_Region

Gender

Canada Foreign Total

M 16 14 30

F 10 22 32

Total 26 36 62

See Principles See Algorithm

Prime

Generalized

Relation

Initial

Relation

See Implementation See Analytical Characterization

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.,

. %] 47 : [ " " ) ( _ %] 53 : [ " " ) ( _

) ( ) (

t foreign x region birth t Canada x region birth

x male x grad

= v =

¬ .

April 30, 2012 14

Presentation—Generalized Relation

April 30, 2012 15

Presentation—Crosstab

April 30, 2012 16

Implementation by Cube Technology

Construct a data cube on-the-fly for the given data

mining query

Facilitate efficient drill-down analysis

May increase the response time

A balanced solution: precomputation of ―subprime‖

relation

Use a predefined & precomputed data cube

Construct a data cube beforehand

Facilitate not only the attribute-oriented induction,

but also attribute relevance analysis, dicing, slicing,

roll-up and drill-down

Cost of cube computation and the nontrivial storage

overhead

April 30, 2012 17

Chapter 5: Concept Description:

Characterization and Comparison

What is concept description?

Data generalization and summarization-based

characterization

Analytical characterization: Analysis of attribute relevance

Mining class comparisons: Discriminating between

different classes

Mining descriptive statistical measures in large databases

Discussion

Summary

April 30, 2012 18

Characterization vs. OLAP

Similarity:

Presentation of data summarization at multiple levels of

abstraction.

Interactive drilling, pivoting, slicing and dicing.

Differences:

Automated desired level allocation.

Dimension relevance analysis and ranking when there

are many relevant dimensions.

Sophisticated typing on dimensions and measures.

Analytical characterization: data dispersion analysis.

April 30, 2012 19

Attribute Relevance Analysis

Why?

Which dimensions should be included?

How high level of generalization?

Automatic vs. interactive

Reduce # attributes; easy to understand patterns

What?

statistical method for preprocessing data

filter out irrelevant or weakly relevant attributes

retain or rank the relevant attributes

relevance related to dimensions and levels

analytical characterization, analytical comparison

April 30, 2012 20

Attribute relevance analysis (cont‘d)

How?

Data Collection

Analytical Generalization

Use information gain analysis (e.g., entropy or other

measures) to identify highly relevant dimensions and levels.

Relevance Analysis

Sort and select the most relevant dimensions and levels.

Attribute-oriented Induction for class description

On selected dimension/level

OLAP operations (e.g. drilling, slicing) on relevance

rules

April 30, 2012 21

Relevance Measures

Quantitative relevance measure determines the

classifying power of an attribute within a set of

data.

Methods

information gain (ID3)

gain ratio (C4.5)

gini index

_

2

contingency table statistics

uncertainty coefficient

April 30, 2012 22

Information-Theoretic Approach

Decision tree

each internal node tests an attribute

each branch corresponds to attribute value

each leaf node assigns a classification

ID3 algorithm

build decision tree based on training objects with

known class labels to classify testing objects

rank attributes with information gain measure

minimal height

the least number of tests to classify an object

See example

April 30, 2012 23

Top-Down Induction of Decision Tree

Attributes = {Outlook, Temperature, Humidity, Wind}

Outlook

Humidity

Wind

sunny rain

overcast

yes

no yes

high

normal

no

strong

weak

yes

PlayTennis = {yes, no}

April 30, 2012 24

Entropy and Information Gain

S contains s

i

tuples of class C

i

for i = {1, …, m}

Information measures info required to classify

any arbitrary tuple

Entropy of attribute A with values {a

1

,a

2

,…,a

v

}

Information gained by branching on attribute A

s

s

log

s

s

) ,...,s ,s s I(

i

m

i

i

m 2 1 2

1

¿

=

÷ =

) s ,..., s ( I

s

s ... s

E(A) mj j

v

j

mj j

1

1

1

¿

=

+ +

=

E(A) ) s ,..., s , I(s Gain(A) m ÷ = 2 1

April 30, 2012 25

Example: Analytical Characterization

Task

Mine general characteristics describing graduate

students using analytical characterization

Given

attributes name, gender, major, birth_place,

birth_date, phone#, and gpa

Gen(a

i

) = concept hierarchies on a

i

U

i

= attribute analytical thresholds for a

i

T

i

= attribute generalization thresholds for a

i

R = attribute relevance threshold

April 30, 2012 26

Example: Analytical

Characterization (cont‘d)

1. Data collection

target class: graduate student

contrasting class: undergraduate student

2. Analytical generalization using U

i

attribute removal

remove name and phone#

attribute generalization

generalize major, birth_place, birth_date and gpa

accumulate counts

candidate relation: gender, major, birth_country,

age_range and gpa

April 30, 2012 27

Example: Analytical characterization (2)

gender major birth_country age_range gpa count

M Science Canada 20-25 Very_good 16

F Science Foreign 25-30 Excellent 22

M Engineering Foreign 25-30 Excellent 18

F Science Foreign 25-30 Excellent 25

M Science Canada 20-25 Excellent 21

F Engineering Canada 20-25 Excellent 18

Candidate relation for Target class: Graduate students (E=120)

gender major birth_country age_range gpa count

M Science Foreign <20 Very_good 18

F Business Canada <20 Fair 20

M Business Canada <20 Fair 22

F Science Canada 20-25 Fair 24

M Engineering Foreign 20-25 Very_good 22

F Engineering Canada <20 Excellent 24

Candidate relation for Contrasting class: Undergraduate students (E=130)

April 30, 2012 28

Example: Analytical characterization (3)

3. Relevance analysis

Calculate expected info required to classify an

arbitrary tuple

Calculate entropy of each attribute: e.g. major

9988 0

250

130

250

130

250

120

250

120

130 120 2 2 2 1 . log log ) , I( ) s , I(s = ÷ ÷ = =

For major=”Science”: S

11

=84 S

21

=42 I(s

11

,s

21

)=0.9183

For major=”Engineering”: S

12

=36 S

22

=46 I(s

12

,s

22

)=0.9892

For major=”Business”: S

13

=0 S

23

=42 I(s

13

,s

23

)=0

Number of grad

students in “Science”

Number of undergrad

students in “Science”

April 30, 2012 29

Example: Analytical Characterization (4)

Calculate expected info required to classify a given

sample if S is partitioned according to the attribute

Calculate information gain for each attribute

Information gain for all attributes

7873 0

250

42

250

82

250

126

23 13 22 12 21 11 . ) s , s ( I ) s , s ( I ) s , s ( I E(major) = + + =

2115 0 2 1 . E(major) ) s , I(s ) Gain(major = ÷ =

Gain(gender) = 0.0003

Gain(birth_country) = 0.0407

Gain(major) = 0.2115

Gain(gpa) = 0.4490

Gain(age_range) = 0.5971

April 30, 2012 30

Example: Analytical characterization (5)

4. Initial working relation (W

0

) derivation

R = 0.1

remove irrelevant/weakly relevant attributes from candidate

relation => drop gender, birth_country

remove contrasting class candidate relation

5. Perform attribute-oriented induction on W

0

using T

i

major age_range gpa count

Science 20-25 Very_good 16

Science 25-30 Excellent 47

Science 20-25 Excellent 21

Engineering 20-25 Excellent 18

Engineering 25-30 Excellent 18

Initial target class working relation W

0

: Graduate students

April 30, 2012 31

Chapter 5: Concept Description:

Characterization and Comparison

What is concept description?

Data generalization and summarization-based

characterization

Analytical characterization: Analysis of attribute relevance

Mining class comparisons: Discriminating between

different classes

Mining descriptive statistical measures in large databases

Discussion

Summary

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.

April 30, 2012 33

Example: Analytical comparison

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

April 30, 2012 34

Example: Analytical comparison (2)

Given

attributes name, gender, major, birth_place,

birth_date, residence, phone# and gpa

Gen(a

i

) = concept hierarchies on attributes a

i

U

i

= attribute analytical thresholds for

attributes a

i

T

i

= attribute generalization thresholds for

attributes a

i

R = attribute relevance threshold

April 30, 2012 35

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

April 30, 2012 36

Example: Analytical comparison (4)

Birth_country Age_range Gpa Count%

Canada 20-25 Good 5.53%

Canada 25-30 Good 2.32%

Canada Over_30 Very_good 5.86%

… … … …

Other Over_30 Excellent 4.68%

Prime generalized relation for the target class: Graduate students

Birth_country Age_range Gpa Count%

Canada 15-20 Fair 5.53%

Canada 15-20 Good 4.53%

… … … …

Canada 25-30 Good 5.02%

… … … …

Other Over_30 Excellent 0.68%

Prime generalized relation for the contrasting class: Undergraduate students

April 30, 2012 37

Example: Analytical comparison (5)

4. Drill down, roll up and other OLAP operations

on target and contrasting classes to adjust

levels of abstractions of resulting description

5. Presentation

as generalized relations, crosstabs, bar

charts, pie charts, or rules

contrasting measures to reflect comparison

between target and contrasting classes

e.g. count%

April 30, 2012 38

Quantitative Discriminant Rules

Cj = target class

q

a

= a generalized tuple covers some tuples of class

but can also cover some tuples of contrasting class

d-weight

range: [0, 1]

quantitative discriminant rule form

¿

=

e

e

= ÷

m

i

i a

j a

) C count(q

) C count(q

weight d

1

d_weight] : [d X) condition( ss(X) target_cla X, : ¬

April 30, 2012 39

Example: Quantitative Discriminant

Rule

Quantitative discriminant rule

where 90/(90+120) = 30%

Status Birth_country Age_range Gpa Count

Graduate Canada 25-30 Good 90

Undergraduate Canada 25-30 Good 210

Count distribution between graduate and undergraduate students for a generalized tuple

%] 30 : [ " " ) ( " 30 25 " ) ( _ " " ) ( _

) ( _ ,

d good X gpa X range age Canada X country birth

X student graduate X

= . ÷ = . =

: ¬

April 30, 2012 40

Class Description

Quantitative characteristic rule

necessary

Quantitative discriminant rule

sufficient

Quantitative description rule

necessary and sufficient

] w : d , w : [t ... ] w : d , w : [t n n 1 1 1

'

v v

'

· ¬

(X) condition (X) condition

ss(X) target_cla X,

n

d_weight] : [d X) condition( ss(X) target_cla X, : ¬

t_weight] : [t X) condition( ss(X) target_cla X, ¬ ¬

April 30, 2012 41

Example: Quantitative Description

Rule

Quantitative description rule for target class Europe

Location/item TV Computer Both_items

Count t-wt d-wt Count t-wt d-wt Count t-wt d-wt

Europe 80 25% 40% 240 75% 30% 320 100% 32%

N_Am 120 17.65% 60% 560 82.35% 70% 680 100% 68%

Both_

regions

200 20% 100% 800 80% 100% 1000 100% 100%

Crosstab showing associated t-weight, d-weight values and total number (in thousands) of TVs and

computers sold at AllElectronics in 1998

30%] : d 75%, : [t 40%] : d 25%, : [t ) computer" " (item(X) ) TV" " (item(X)

Europe(X) X,

= v =

· ¬

April 30, 2012 42

Chapter 5: Concept Description:

Characterization and Comparison

What is concept description?

Data generalization and summarization-based

characterization

Analytical characterization: Analysis of attribute relevance

Mining class comparisons: Discriminating between

different classes

Mining descriptive statistical measures in large databases

Discussion

Summary

April 30, 2012 43

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

April 30, 2012 44

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:

¿

=

=

n

i

i

x

n

x

1

1

¿

¿

=

=

=

n

i

i

n

i

i i

w

x w

x

1

1

c

f

l f n

L median

median

)

) ( 2 /

(

1

¿

÷

+ =

) ( 3 median mean mode mean ÷ × = ÷

April 30, 2012 45

Measuring the Dispersion of Data

Quartiles, outliers and boxplots

Quartiles: Q

1

(25

th

percentile), Q

3

(75

th

percentile)

Inter-quartile range: IQR = Q

3

–

Q

1

Five number summary: min, Q

1

, M,

Q

3

, 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 s

2

: (algebraic, scalable computation)

Standard deviation s is the square root of variance s

2

¿ ¿ ¿

= = =

÷

÷

= ÷

÷

=

n

i

n

i

i i

n

i

i

x

n

x

n

x x

n

s

1 1

2 2

1

2 2

] ) (

1

[

1

1

) (

1

1

April 30, 2012 46

Boxplot Analysis

Five-number summary of a distribution:

Minimum, Q1, M, Q3, Maximum

Boxplot

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

April 30, 2012 47

A Boxplot

A boxplot

April 30, 2012 48

Visualization of Data

Dispersion: Boxplot Analysis

April 30, 2012 49

Mining Descriptive Statistical Measures in

Large Databases

Variance

Standard deviation: the square root of the variance

Measures spread about the mean

It is zero if and only if all the values are equal

Both the deviation and the variance are algebraic

( )

(

¸

(

¸

÷

÷

= ÷

÷

=

¿ ¿ ¿

=

2

2

1

2 2

1

1

1

) (

1

1

i i

n

i

i

x

n

x

n

x x

n

s

April 30, 2012 50

Histogram Analysis

Graph displays of basic statistical class descriptions

Frequency histograms

A univariate graphical method

Consists of a set of rectangles that reflect the counts or

frequencies of the classes present in the given data

April 30, 2012 51

Quantile Plot

Displays all of the data (allowing the user to assess

both the overall behavior and unusual occurrences)

Plots quantile information

For a data x

i

data sorted in increasing order, f

i

indicates that approximately 100 f

i

% of the data are

below or equal to the value x

i

April 30, 2012 52

Quantile-Quantile (Q-Q) Plot

Graphs the quantiles of one univariate distribution

against the corresponding quantiles of another

Allows the user to view whether there is a shift in going

from one distribution to another

April 30, 2012 53

Scatter plot

Provides a first look at bivariate data to see clusters of

points, outliers, etc

Each pair of values is treated as a pair of coordinates

and plotted as points in the plane

April 30, 2012 54

Loess Curve

Adds a smooth curve to a scatter plot in order to

provide better perception of the pattern of dependence

Loess curve is fitted by setting two parameters: a

smoothing parameter, and the degree of the

polynomials that are fitted by the regression

April 30, 2012 55

Graphic Displays of Basic Statistical

Descriptions

Histogram: (shown before)

Boxplot: (covered before)

Quantile plot: each value x

i

is paired with f

i

indicating

that approximately 100 f

i

% of data are s x

i

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

April 30, 2012 56

Chapter 5: Concept Description:

Characterization and Comparison

What is concept description?

Data generalization and summarization-based

characterization

Analytical characterization: Analysis of attribute relevance

Mining class comparisons: Discriminating between

different classes

Mining descriptive statistical measures in large databases

Discussion

Summary

April 30, 2012 57

AO Induction vs. Learning-from-

example Paradigm

Difference in philosophies and basic assumptions

Positive and negative samples in learning-from-

example: positive used for generalization, negative -

for specialization

Positive samples only in data mining: hence

generalization-based, to drill-down backtrack the

generalization to a previous state

Difference in methods of generalizations

Machine learning generalizes on a tuple by tuple basis

Data mining generalizes on an attribute by attribute

basis

April 30, 2012 58

Comparison of Entire vs. Factored

Version Space

April 30, 2012 59

Incremental and Parallel Mining of

Concept Description

Incremental mining: revision based on newly added data

ADB

Generalize ADB to the same level of abstraction in the

generalized relation R to derive AR

Union R U AR, i.e., merge counts and other statistical

information to produce a new relation R‘

Similar philosophy can be applied to data sampling,

parallel and/or distributed mining, etc.

April 30, 2012 60

Chapter 5: Concept Description:

Characterization and Comparison

What is concept description?

Data generalization and summarization-based

characterization

Analytical characterization: Analysis of attribute relevance

Mining class comparisons: Discriminating between

different classes

Mining descriptive statistical measures in large databases

Discussion

Summary

April 30, 2012 61

Summary

Concept description: characterization and discrimination

OLAP-based vs. attribute-oriented induction

Efficient implementation of AOI

Analytical characterization and comparison

Mining descriptive statistical measures in large

databases

Discussion

Incremental and parallel mining of description

Descriptive mining of complex types of data

April 30, 2012 62

References

Y. Cai, N. Cercone, and J. Han. Attribute-oriented induction in relational databases. In

G. Piatetsky-Shapiro and W. J. Frawley, editors, Knowledge Discovery in Databases,

pages 213-228. AAAI/MIT Press, 1991.

S. Chaudhuri and U. Dayal. An overview of data warehousing and OLAP technology.

ACM SIGMOD Record, 26:65-74, 1997

C. Carter and H. Hamilton. Efficient attribute-oriented generalization for knowledge

discovery from large databases. IEEE Trans. Knowledge and Data Engineering,

10:193-208, 1998.

W. Cleveland. Visualizing Data. Hobart Press, Summit NJ, 1993.

J. L. Devore. Probability and Statistics for Engineering and the Science, 4th ed.

Duxbury Press, 1995.

T. G. Dietterich and R. S. Michalski. A comparative review of selected methods for

learning from examples. In Michalski et al., editor, Machine Learning: An Artificial

Intelligence Approach, Vol. 1, pages 41-82. Morgan Kaufmann, 1983.

J. Gray, S. Chaudhuri, A. Bosworth, A. Layman, D. Reichart, M. Venkatrao, F. Pellow,

and H. Pirahesh. Data cube: A relational aggregation operator generalizing group-by,

cross-tab and sub-totals. Data Mining and Knowledge Discovery, 1:29-54, 1997.

J. Han, Y. Cai, and N. Cercone. Data-driven discovery of quantitative rules in

relational databases. IEEE Trans. Knowledge and Data Engineering, 5:29-40, 1993.

April 30, 2012 63

References (cont.)

J. Han and Y. Fu. Exploration of the power of attribute-oriented induction in data

mining. In U.M. Fayyad, G. Piatetsky-Shapiro, P. Smyth, and R. Uthurusamy, editors,

Advances in Knowledge Discovery and Data Mining, pages 399-421. AAAI/MIT Press,

1996.

R. A. Johnson and D. A. Wichern. Applied Multivariate Statistical Analysis, 3rd ed.

Prentice Hall, 1992.

E. Knorr and R. Ng. Algorithms for mining distance-based outliers in large datasets.

VLDB'98, New York, NY, Aug. 1998.

H. Liu and H. Motoda. Feature Selection for Knowledge Discovery and Data Mining.

Kluwer Academic Publishers, 1998.

R. S. Michalski. A theory and methodology of inductive learning. In Michalski et al.,

editor, Machine Learning: An Artificial Intelligence Approach, Vol. 1, Morgan

Kaufmann, 1983.

T. M. Mitchell. Version spaces: A candidate elimination approach to rule learning.

IJCAI'97, Cambridge, MA.

T. M. Mitchell. Generalization as search. Artificial Intelligence, 18:203-226, 1982.

T. M. Mitchell. Machine Learning. McGraw Hill, 1997.

J. R. Quinlan. Induction of decision trees. Machine Learning, 1:81-106, 1986.

D. Subramanian and J. Feigenbaum. Factorization in experiment generation. AAAI'86,

Philadelphia, PA, Aug. 1986.

April 30, 2012 64

http://www.cs.sfu.ca/~han/dmbook

Thank you !!!

**Chapter 5: Concept Description: Characterization and Comparison
**

What is concept description? Data generalization and summarization-based characterization Analytical characterization: Analysis of attribute relevance Mining class comparisons: Discriminating between different classes Mining descriptive statistical measures in large databases Discussion Summary

2

April 30, 2012

**What is Concept 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. 2012 . 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 4 April 30.

Chapter 5: Concept Description: Characterization and Comparison What is concept description? Data generalization and summarization-based characterization Analytical characterization: Analysis of attribute relevance Mining class comparisons: Discriminating between different classes Mining descriptive statistical measures in large databases Discussion Summary 5 April 30. 2012 .

Data Generalization and Summarizationbased 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. 1 2 3 4 Conceptual levels Approaches: Data cube approach(OLAP approach) Attribute-oriented induction approach 6 5 April 30. 2012 .

sum( ).g.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. Lack of intelligent analysis. average( ).. 2012 . can‘t tell which dimensions should be used and what levels should the generalization reach 7 Limitations April 30. max( ) Generalization and specialization can be performed on a data cube by roll-up and drill-down handle only dimensions of simple nonnumeric data and measures of simple aggregated numeric values. count( ).

**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.

April 30, 2012

8

**Basic Principles of AttributeOriented 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. see example

**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. Presentation: User interaction: (1) adjust levels by drilling, (2) pivoting, (3) mapping into rules, cross tabs, visualization presentations.

See example See complexity

See Implementation

gpa from student where status in {―Msc‖. phone#.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. major. birth_place. birth_date. birth_place. birth_date. 2012 . gender. phone#. residence. residence. ―MBA‖. gpa from student where status in ―graduate‖ Corresponding SQL statement: Select name. gender. major. ―PhD‖ } 11 April 30.

. USA 25-8-70 … … … Sci.83 … Excl. 28-7-75 Canada Physics Seattle.70 3. VG.Class Characterization: An Example Name Jim Initial Woodman Relation Scott Lachance Laura Lee … Removed Gender M M F … Retained Major CS Birth-Place Birth_date Residence 3511 Main St. Richmond 345 1st Ave. WA. 8-12-76 Canada CS Montreal... Que. Bus Country Age range Age_range 20-25 25-30 … City GPA Very-good Excellent … Gender Major Birth_region Canada Foreign … Residence Richmond Burnaby … Count 16 22 … Prime Generalized Relation M F … Science Science … Birth_Region Canada Gender M F Total See Principles See Algorithm See Implementation Foreign 14 22 36 Total 30 32 62 16 10 26 See Analytical Characterization ..Eng..67 3. Vancouver. Burnaby … Phone # 687-4598 253-9106 420-5232 … Removed GPA 3. Richmond 125 Austin Ave.BC.

Presentation of Generalized Results Generalized relation: Relations where some or all attributes are generalized. curves. Mapping results into cross tabulation form (similar to contingency tables). with counts or other aggregation values accumulated. .. e. Cross tabulation: Quantitative characteristic rules: grad( x) male( x) birth _ region( x) "Canada"[t :53%] birth _ region( x) " foreign"[t : 47%]. and other visual forms. cubes. Mapping generalized result into characteristic rules with quantitative information associated with it. bar charts.g. Visualization techniques: Pie charts.

2012 14 .Presentation—Generalized Relation April 30.

Presentation—Crosstab April 30. 2012 15 .

dicing. but also attribute relevance analysis. roll-up and drill-down Cost of cube computation and the nontrivial storage overhead 16 April 30. 2012 . slicing.Implementation by Cube Technology Construct a data cube on-the-fly for the given data mining query Facilitate efficient drill-down analysis May increase the response time A balanced solution: precomputation of ―subprime‖ relation Use a predefined & precomputed data cube Construct a data cube beforehand Facilitate not only the attribute-oriented induction.

2012 .Chapter 5: Concept Description: Characterization and Comparison What is concept description? Data generalization and summarization-based characterization Analytical characterization: Analysis of attribute relevance Mining class comparisons: Discriminating between different classes Mining descriptive statistical measures in large databases Discussion Summary 17 April 30.

Sophisticated typing on dimensions and measures.Characterization vs. OLAP Similarity: Presentation of data summarization at multiple levels of abstraction. Automated desired level allocation. pivoting. Interactive drilling. 18 Differences: April 30. slicing and dicing. Dimension relevance analysis and ranking when there are many relevant dimensions. Analytical characterization: data dispersion analysis. 2012 .

easy to understand patterns What? statistical method for preprocessing data filter out irrelevant or weakly relevant attributes retain or rank the relevant attributes April 30. interactive Reduce # attributes. analytical comparison 19 .Attribute Relevance Analysis Why? Which dimensions should be included? How high level of generalization? Automatic vs. 2012 relevance related to dimensions and levels analytical characterization.

. entropy or other measures) to identify highly relevant dimensions and levels. 2012 . Relevance Analysis Attribute-oriented Induction for class description On selected dimension/level OLAP operations (e. Sort and select the most relevant dimensions and levels. slicing) on relevance rules 20 April 30.Attribute relevance analysis (cont‘d) How? Data Collection Analytical Generalization Use information gain analysis (e.g.g. drilling.

Relevance Measures Quantitative relevance measure determines the classifying power of an attribute within a set of data. Methods information gain (ID3) gain ratio (C4. 2012 .5) gini index 2 contingency table statistics uncertainty coefficient 21 April 30.

2012 22 .Information-Theoretic Approach Decision tree each internal node tests an attribute each branch corresponds to attribute value each leaf node assigns a classification ID3 algorithm build decision tree based on training objects with known class labels to classify testing objects rank attributes with information gain measure minimal height the least number of tests to classify an object See example April 30.

Wind} PlayTennis = {yes. Temperature. 2012 normal yes strong no weak yes 23 . Humidity. no} Outlook sunny Humidity overcast yes rain Wind high no April 30.Top-Down Induction of Decision Tree Attributes = {Outlook.

.a2... m} Information measures info required to classify any arbitrary tuple s s I( s .. s 2 . smj I ( s1 j . 2012 24 . smj ) s j 1 v Information gained by branching on attribute A Gain(A) I(s1..av} E(A) s1 j ......….s .s ) log s s m i i 1 2 m 2 i 1 Entropy of attribute A with values {a1. …...sm) E(A) April 30.Entropy and Information Gain S contains si tuples of class Ci for i = {1...

major. 2012 . gender. birth_place.Example: Analytical Characterization Task Mine general characteristics describing graduate students using analytical characterization Given attributes name. phone#. and gpa Gen(ai) = concept hierarchies on ai Ui = attribute analytical thresholds for ai Ti = attribute generalization thresholds for ai R = attribute relevance threshold 25 April 30. birth_date.

birth_date and gpa accumulate counts attribute generalization candidate relation: gender. birth_country. 2012 26 . Data collection target class: graduate student contrasting class: undergraduate student 2. birth_place. major. Analytical generalization using Ui attribute removal remove name and phone# generalize major.Example: Analytical Characterization (cont‘d) 1. age_range and gpa April 30.

Example: Analytical characterization (2) gender major birth_country age_range gpa count M F M F M F Science Science Engineering Science Science Engineering Canada Foreign Foreign Foreign Canada Canada 20-25 25-30 25-30 25-30 20-25 20-25 Very_good Excellent Excellent Excellent Excellent Excellent 16 22 18 25 21 18 Candidate relation for Target class: Graduate students (=120) gender major birth_country age_range gpa count M F M F M F Science Business Business Science Engineering Engineering Foreign Canada Canada Canada Foreign Canada <20 <20 <20 20-25 20-25 <20 Very_good Fair Fair Fair Very_good Excellent 18 20 22 24 22 24 27 Candidate relation for Contrasting class: Undergraduate students (=130) April 30. 2012 .

2012 .130 ) 120 120 130 130 log 2 log 2 0.s23)=0 Number of undergrad students in “Science” 28 For major=”Engineering”: S12=36 S13=0 Number of grad students in “Science” April 30.9892 I(s13.s22)=0. Relevance analysis Calculate expected info required to classify an arbitrary tuple I(s1.g.s21)=0.9988 250 250 250 250 Calculate entropy of each attribute: e.9183 I(s12. major For major=”Science”: For major=”Business”: S11=84 S21=42 S22=46 S23=42 I(s11. s 2 ) I(120.Example: Analytical characterization (3) 3.

s 23 ) 0.7873 250 250 250 Calculate information gain for each attribute Gain(major) I(s1.2115 Information gain for all attributes Gain(gender) Gain(birth_country) Gain(major) Gain(gpa) Gain(age_range) = 0. s 2 ) E(major) 0.4490 = 0. s 22 ) I ( s13.0003 = 0.5971 29 April 30. 2012 . s 21 ) I ( s12 .0407 = 0.2115 = 0.Example: Analytical Characterization (4) Calculate expected info required to classify a given sample if S is partitioned according to the attribute E(major) 126 82 42 I ( s11.

birth_country remove contrasting class candidate relation major Science Science Science Engineering Engineering age_range 20-25 25-30 20-25 20-25 25-30 gpa Very_good Excellent Excellent Excellent Excellent count 16 47 21 18 18 Initial target class working relation W0: Graduate students 5. Perform attribute-oriented induction on W0 using Ti 30 April 30. 2012 .Example: Analytical characterization (5) 4. Initial working relation (W0) derivation R = 0.1 remove irrelevant/weakly relevant attributes from candidate relation => drop gender.

2012 .Chapter 5: Concept Description: Characterization and Comparison What is concept description? Data generalization and summarization-based characterization Analytical characterization: Analysis of attribute relevance Mining class comparisons: Discriminating between different classes Mining descriptive statistical measures in large databases Discussion Summary 31 April 30.

. 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 .Mining Class Comparisons Comparison: Comparing two or more classes.distribution between classes Highlight the tuples with strong discriminant features Relevance Analysis: Find attributes (features) which best distinguish different classes.

gender. residence. DMQL query use Big_University_DB mine comparison as “grad_vs_undergrad_students” in relevance to name. birth_date.Example: Analytical comparison Task Compare graduate and undergraduate students using discriminant rule. birth_place. gpa for “graduate_students” where status in “graduate” versus “undergraduate_students” where status in “undergraduate” analyze count% from student April 30. 2012 33 . major. phone#.

birth_place.Example: Analytical comparison (2) Given attributes name. birth_date. 2012 . gender. 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 34 April 30. major.

Example: Analytical comparison (3) 1. phone# 3. 2012 . major. Data collection target and contrasting classes 2. gender. Attribute relevance analysis remove attributes name. Synchronous generalization controlled by user-specified dimension thresholds prime target and contrasting class(es) relations/cuboids 35 April 30.

68% Prime generalized relation for the contrasting class: Undergraduate students April 30.32% 5.53% 2.68% Prime generalized relation for the target class: Graduate students Birth_country Canada Canada … Canada … Other Age_range 15-20 15-20 … 25-30 … Over_30 Gpa Fair Good … Good … Excellent Count% 5. 2012 36 .02% … 0.86% … 4.Example: Analytical comparison (4) Birth_country Canada Canada Canada … Other Age_range 20-25 25-30 Over_30 … Over_30 Gpa Good Good Very_good … Excellent Count% 5.53% 4.53% … 5.

count% 37 April 30.g.Example: Analytical comparison (5) 4. Presentation as generalized relations. 2012 . Drill down. crosstabs. pie charts. bar charts. roll up and other OLAP operations on target and contrasting classes to adjust levels of abstractions of resulting description 5. or rules contrasting measures to reflect comparison between target and contrasting classes e.

2012 38 .Quantitative Discriminant Rules Cj = target class qa = a generalized tuple covers some tuples of class but can also cover some tuples of contrasting class d-weight range: [0. target_class(X) condition(X) [d : d_weight] April 30. 1] d weight count(qa Cj ) a count(q C ) i i 1 m quantitative discriminant rule form X.

2012 39 . graduate _ student( X ) birth _ country( X ) " Canada" age _ range( X ) "25 30" gpa( X ) " good" [d : 30 %] where 90/(90+120) = 30% April 30.Example: Quantitative Discriminant Rule Status Graduate Undergraduate Birth_country Canada Canada Age_range 25-30 25-30 Gpa Good Good Count 90 210 Count distribution between graduate and undergraduate students for a generalized tuple Quantitative discriminant rule X .

target_class(X) condition(X) [d : d_weight] sufficient Quantitative description rule X. target_class(X) condition1(X) [t : w1. 2012 40 . d : w 1] ... target_class(X) condition(X) [t : t_weight] necessary Quantitative discriminant rule X. conditionn(X) [t : wn.Class Description Quantitative characteristic rule X. d : w n] necessary and sufficient April 30.

65% 20% d-wt 40% 60% 100% Count 240 560 800 Computer t-wt 75% 82.35% 80% d-wt 30% 70% 100% Count 320 680 1000 Both_items t-wt 100% 100% 100% d-wt 32% 68% 100% Crosstab showing associated t-weight. d : 30%] 41 April 30. 2012 . d : 40%] (item(X) " computer" ) [t : 75%. Europe(X) (item(X) " TV" ) [t : 25%.Example: Quantitative Description Rule Location/item Count Europe N_Am Both_ regions 80 120 200 TV t-wt 25% 17. d-weight values and total number (in thousands) of TVs and computers sold at AllElectronics in 1998 Quantitative description rule for target class Europe X.

2012 .Chapter 5: Concept Description: Characterization and Comparison What is concept description? Data generalization and summarization-based characterization Analytical characterization: Analysis of attribute relevance Mining class comparisons: Discriminating between different classes Mining descriptive statistical measures in large databases Discussion Summary 42 April 30.

etc. 2012 . variance. max.Mining Data Dispersion Characteristics Motivation To better understand the data: central tendency. Data dispersion: analyzed with multiple granularities of precision Boxplot or quantile analysis on sorted intervals Folding measures into numerical dimensions Boxplot or quantile analysis on the transformed cube 43 Data dispersion characteristics Numerical dimensions correspond to sorted intervals Dispersion analysis on computed measures April 30. quantiles. variation and spread median. min. outliers.

2012 . or average of the middle two values otherwise estimated by interpolation median L1 ( n / 2 ( f )l f median )c Mode Value that occurs most frequently in the data Unimodal. bimodal. trimodal Empirical formula: mean mode 3 (mean median) 44 April 30.Measuring the Central Tendency Mean Weighted arithmetic mean 1 n x xi n i 1 x w x i 1 n i n i Median: A holistic measure w i 1 i Middle value if odd number of values.

Measuring the Dispersion of Data Quartiles. a value higher/lower than 1. scalable computation) s 2 n n n 1 1 1 2 2 ( xi x ) n 1 [ xi n ( xi ) 2 ] n 1 i 1 i 1 i 1 April 30. 2012 Standard deviation s is the square root of variance s2 45 . max Boxplot: ends of the box are the quartiles. whiskers. Q3 (75th percentile) Inter-quartile range: IQR = Q3 – Q1 Five number summary: min. median is marked. Q3.5 x IQR Variance and standard deviation Variance s2: (algebraic. and plot outlier individually Outlier: usually. Q1. outliers and boxplots Quartiles: Q1 (25th percentile). M.

Q1. Maximum Boxplot Data is represented with a box The ends of the box are at the first and third quartiles.e. i. 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 46 April 30.Boxplot Analysis Five-number summary of a distribution: Minimum. Q3. 2012 .. M.

2012 47 .A Boxplot A boxplot April 30.

2012 48 .Visualization of Data Dispersion: Boxplot Analysis April 30.

2012 49 .Mining Descriptive Statistical Measures in Large Databases Variance 1 n 1 1 2 2 s ( xi x ) 2 xi2 xi n 1 i 1 n 1 n Standard deviation: the square root of the variance Measures spread about the mean It is zero if and only if all the values are equal Both the deviation and the variance are algebraic April 30.

2012 50 .Histogram Analysis Graph displays of basic statistical class descriptions Frequency histograms A univariate graphical method Consists of a set of rectangles that reflect the counts or frequencies of the classes present in the given data April 30.

2012 51 .Quantile Plot Displays all of the data (allowing the user to assess both the overall behavior and unusual occurrences) Plots quantile information For a data xi data sorted in increasing order. fi indicates that approximately 100 fi% of the data are below or equal to the value xi April 30.

2012 52 .Quantile-Quantile (Q-Q) Plot Graphs the quantiles of one univariate distribution against the corresponding quantiles of another Allows the user to view whether there is a shift in going from one distribution to another April 30.

Scatter plot Provides a first look at bivariate data to see clusters of points. etc Each pair of values is treated as a pair of coordinates and plotted as points in the plane April 30. outliers. 2012 53 .

and the degree of the polynomials that are fitted by the regression April 30.Loess Curve Adds a smooth curve to a scatter plot in order to provide better perception of the pattern of dependence Loess curve is fitted by setting two parameters: a smoothing parameter. 2012 54 .

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 April 30. 2012 55 .

Chapter 5: Concept Description: Characterization and Comparison What is concept description? Data generalization and summarization-based characterization Analytical characterization: Analysis of attribute relevance Mining class comparisons: Discriminating between different classes Mining descriptive statistical measures in large databases Discussion Summary 56 April 30. 2012 .

2012 . Learning-fromexample Paradigm Difference in philosophies and basic assumptions Positive and negative samples in learning-fromexample: positive used for generalization. to drill-down backtrack the generalization to a previous state Difference in methods of generalizations Machine learning generalizes on a tuple by tuple basis Data mining generalizes on an attribute by attribute basis 57 April 30.AO Induction vs. negative for specialization Positive samples only in data mining: hence generalization-based.

Comparison of Entire vs. Factored Version Space April 30. 2012 58 .

e. i.Incremental and Parallel Mining of Concept Description Incremental mining: revision based on newly added data DB Generalize DB to the same level of abstraction in the generalized relation R to derive R Union R U R. merge counts and other statistical information to produce a new relation R‘ Similar philosophy can be applied to data sampling.. 2012 59 . etc. April 30. parallel and/or distributed mining.

2012 .Chapter 5: Concept Description: Characterization and Comparison What is concept description? Data generalization and summarization-based characterization Analytical characterization: Analysis of attribute relevance Mining class comparisons: Discriminating between different classes Mining descriptive statistical measures in large databases Discussion Summary 60 April 30.

2012 .Summary Concept description: characterization and discrimination OLAP-based vs. attribute-oriented induction Efficient implementation of AOI Analytical characterization and comparison Mining descriptive statistical measures in large databases Discussion Incremental and parallel mining of description Descriptive mining of complex types of data 61 April 30.

5:29-40. Vol. 4th ed. Hamilton. Pellow. Data cube: A relational aggregation operator generalizing group-by. pages 41-82. Knowledge Discovery in Databases. Knowledge and Data Engineering. S. 2012 . A. Efficient attribute-oriented generalization for knowledge discovery from large databases. Morgan Kaufmann. 1. Knowledge and Data Engineering. Pirahesh. Frawley. M. Bosworth. 1991. Duxbury Press. editors. J. cross-tab and sub-totals. F. Layman. Cai. Cercone. and N. Cleveland. Data Mining and Knowledge Discovery. Cercone. Machine Learning: An Artificial Intelligence Approach. N. Cai. J. and H. 1983. Chaudhuri. A comparative review of selected methods for learning from examples. J. G. Gray. Michalski. pages 213-228. ACM SIGMOD Record. and J. 1993. AAAI/MIT Press. W.. 62 April 30. 1997. S. S. Data-driven discovery of quantitative rules in relational databases. L. IEEE Trans. Summit NJ. Carter and H. Han. editor. 10:193-208. Venkatrao. 1997 C. Piatetsky-Shapiro and W. 1:29-54. IEEE Trans. Dietterich and R. 26:65-74. 1998. An overview of data warehousing and OLAP technology. Devore. Attribute-oriented induction in relational databases. Chaudhuri and U. T. Dayal. Visualizing Data. In Michalski et al. Reichart. Hobart Press. A. D. Han. 1993. Y. J.References Y. 1995. Probability and Statistics for Engineering and the Science. In G.

Motoda.M. Smyth. M. T. Johnson and D. Knorr and R. Philadelphia. Ng. Piatetsky-Shapiro. Exploration of the power of attribute-oriented induction in data mining. P. Generalization as search. Prentice Hall. and R. Quinlan. 1:81-106. D. Aug. M. Version spaces: A candidate elimination approach to rule learning. Fu. J. Artificial Intelligence. Machine Learning: An Artificial Intelligence Approach. Fayyad. 1997. VLDB'98. MA. editors. Mitchell. editor. Applied Multivariate Statistical Analysis. Mitchell. H.) J. 1992. IJCAI'97. T. 1986. R. In U. PA. 3rd ed. Algorithms for mining distance-based outliers in large datasets. In Michalski et al.. A theory and methodology of inductive learning. Feigenbaum. 1. Induction of decision trees. Advances in Knowledge Discovery and Data Mining. Michalski. Kluwer Academic Publishers. A. Subramanian and J. NY. 1998. 2012 . 1996. 1998.References (cont. R. E. pages 399-421. Liu and H. Morgan Kaufmann. McGraw Hill. Cambridge. Uthurusamy. 63 April 30. Feature Selection for Knowledge Discovery and Data Mining. A. AAAI'86. AAAI/MIT Press. G. Vol. M. Han and Y. S. 1986. Machine Learning. 1982. Machine Learning. Mitchell. T. R. Wichern. Aug. Factorization in experiment generation. 1983. 18:203-226. New York.

ca/~han/dmbook Thank you !!! April 30.sfu.cs.http://www. 2012 64 .

03.9.9:65 647.::: .43306507.73:.:..693..66946903..9.:::.: 647.03.51.65.065.:: : 59.6.9.::647..503.6.65.553.9:65: 647.:9: :7769.:: .73..61 #.4331:097.65: #9:5.1.6 4.5.9..51.699.:.3/6.5..95.:1:097.:.:.

5:5303.9/.65.9:65. 1:.:: 0647.

9: 0/:.::: 3..65/.03.9: %3.9651:0945....505.3:: 51.5: 195.503. 1:..5..:.: .9/.9/.73:.1:..::: .

:2 647.0 -79*/.078:8 :3/077.3.93 $89 :80 *&3.3.20 03/07 2./:.0.9:83 7.1.47 -79*5.15.425.4735.90 708/03./:.: :51:0945.30647..0789* 230.8*:3/077.515199.90 ..5./:../:.90*89:/0398 070 89.0 5430 5.0 .8 7.99.9:83 :3/077./*.90*89:/0398 070 89..1./*89:/0398 3700.. 147 7.9:65 '..3.4:39 1742 89:/039 !64/9 .90 .7843.:.094 3.

06507.69/9.0..59.3..:5.65.93.9/..0 /9.9/.0. % ..50.9/.9:631:69 .73.:. .1... ' .:..3.9/.9/.5..3.:..517..9:65 5 . 5 .9/..45194...9..3.9:150765 .90:65.4735....9/..9:631 !64/9 .30647..9:631:69 ..

4735..50.::: .30647.93.9/.96331/:9.503.65 .3:: 946.45194.51065.65 065.:5..69765 &509656:59.:.3..9.3....5.9:65 ..0.9/..06330.9..

:701145:65.503.:: : 93..9:631: 794.9.9..:.51065..65:0/61: !64/9 .

07* 44/ .07* . .0./.0/70.3.3..88&3/077./. 44/ ..30647.4:397 0*7.887.4397.3./.9:65 79*.943147909./:. . 907 ..3.94314790.709.30 5.4735. ./.893.7 . 44/ . .3.3./:.0039 4:39 !7200307.30 5.0/70.0039 4:39 !7200307. 44/ ./.9089:/0398 79*.4:397 0*7./.3.9089:/0398 !64/9 .07* '07*44/ 907 .

0./:/.6930.9.:.51:097.3193. !64/9 .3..:6993: 065.9:65 /.51065.0647.:70.9../:.51065....54. 33:6.9"#679.:::.65:096::.9:65 93316596337.:59.516.1:.30647.5.9.9.0...::: 065.6.65: 65.65 .9.503.503..65:69:3.:9:..9.9.:..:.65 #9:5.9.9 0.4735.

:: 8.03.5.:..5.:: 1.5.3:6069:64..59.:0945.9.0.31.%3: .9..73:6065.:: /.73:603.$. .73069::64.503..

93694 9..5(* / 09 . 9.1:0945...709*.4:39 6..43/943 / /*09( !64/9 .88 .5.5.. Z 8. Z 2 .4:39 6.

./:.0307./:.. 0*7.90./..5.90 &3/077.90 79*.0/9:50 $./:.93 7./.3/:3/077. ..30 5./:.3.3./.:0945./:.4:397 . 44/ / ( 9 !64/9 .9:8 7.. 44/ 44/ 4:39 4:39/897-:943-090037.5.1:0945.5.90 * 89:/039 -79 * .30 5..473$.0 * 7.5.3.. %3 $9.9089:/0398147..4:397 .

.5..709*..3.88 .1:0945.5.88 .5.709*...9:.709*.0.88 ..43/943 9 9*09( 50::..43/9433 9 3 / @ ( 3( 50::.1:097.65 $... $.9 $.51:05.:::097.9. !64/9 ...0.43/943 / /*09( :05.5.9.43/943 9 / @ .093 9.6593 9..93 9.

..65 %3 4..:097..473$..5.943.

90..3/949.902 :7450 *2 49* 70438 %' / 9 425:907 / 9 49*9028 / 9 4:39 9 9 4:39 9 9 4:39 9 9 74889.884.:08..9743.1:097.83 $.::967 :7450 902 %' 9 / ( 902 .5.425:907 9 / ( !64/9 .659369.03.90/9 09 / 09.3/ ..9.3/8 41%'8.3:2-07 394:8..-843.425:907884/...

51647.9...65 ..51:44.96507.:097.:06507..65.59..7.1:097.9.65 .9..9:65 .65..65.3.0.

30.5 195.9.34.9:65::0945.9../..::: 551:097.655...:1 0.3::6./.:.:: :0::65 &44..50 5503.0..9/.0.::0647.93.65 5..:...0.9.0.91.03.5/.3.:9:53.9 !64/9 .9.

15.3145:65: 0699:7651.5.6:69.3::65:69.05..9:..5.3::650647.5..5150.3145:65: 6736.3.50.51.3:6.739.0: 41..698..54.698.3.43.55.0: 6.458.9:.9.65490.14.9.3::65.9519:.5:69410/ !64/9 .9.1...65 .1:79:65.9.9.31.9.0.:6 790:65 6736.:9:5..3: :79:65.53.9.:9: 63154..:79:65.3.5.9.1:79:650.5.65 '6/.15..9...51:79..1 .5.5.39:..3: .0.0 !490.

6.9: :.361154/96.3:69.563:.4.6.5 1.4. 20/..:9 4113.5..9..3:6.9763.65 61 3 3 3 3 113.1/5.5 .:95.3 3.04.3'5150 .04.9.1.9.

3/461..3 1 .6009:46:.3 !64/9 .9461..3 4790.3.3 24/0 L 20.. .1.985.3. 20.3 20/..35. 1 20/.36943. 5461.

:95. 7905.3 5.:79:656.3 $ . $.3:$ .51/6736.9...9.39:. 7905.3:6.9..: $.

9.51..50 : .39369.50: !64/9 .3.9.65 8 3 3 3 3 3 3 ( &.33..8.5 $% .911..6.0:0.65 .50.65 ::.6. 6736..39:.9966.51:6..911.33 "..:8.51736..921 :29:.39./6.5$%$ $ 54/9:44.8.3/9.5:4.9..3:41.945$ $4.9.39511.51:.9./30647.51.9.

5.6736.3:: .

51.54/9:44.921/. .91 8.44 !64/9 .1./6:%$ '41.5:4.635:6...:979:5.6..6 544..9../6 '51:6.51 .9/.35.9.3: ./6.44 6736.1:..65 544$ $ .9 6.9:.5../6. /6 :29:.:1.51 .

6736. -4549 !64/9 .

.5.:.656.3:: !64/9 ...3. :79:656736.

1.&.9./.51.3 ./6..3 6.:.:: ..9..911.3/9...:9::79.5 .9.6.50 .9..0 !64/9 .65.65.9.98.33.:8....9966.4.:9:5 .0.50...51.55:097.50 3 8 3 3 3 &.51653..1.:96.3:.

:6/..71:73.03..:::79:5.:.4: 5.065..9..:69 9850:6.9.45.69.5.:. !64/9 .::1:097..69.0.:6...303.690.34..65: 9850:..3:: 9.61 65::.:.70.:0:.930.51.53:.

..3 6.69.:.:9.1..5.1.9 /36698.69..:56919 510.65 69.....:.:8..77964. .336.36009950: #36.3 5694.5.$...3.33/..3365.3 !64/9 ...1.6...515:.:::: /6.:69. 1..6.3#36. :73.15509.

$.3.5.

$.5.3 $.

9.$ #36.8.9/.6.565 964651:.1:.5:.9 !64/9 .9.5..3:6655.:.65.65 ..7:.9 336:.56.3:6..5.6.0699:765158.9.9:.9/.:9.56.

..0 .7...9736.9.1.96.39:..96066915.9..:5.07.:6.: .&0.5 !64/9 .1..:.9:. #961:.1.73../.3662.:765.3::.51736...6:03:.9:6 765.

656.7.956175150 6::09:.9.6..56919..5.1/..51.3:..67.6 7961/..9.4.4.9.1/:.. 763564....99::65 !64/9 .196.:466.6::9 11:.9.57. :466..09.97907.:0..9:.9736.

3..4 :65/69 6736..5.5 .3736.:0&...91.9..69..0..65: :.3 :097.9A $.:.5..3 :7.0.:6.. 510.70:73...3 61.77964. 0691/69 $.

8.5.3 8.

5..9736.8 736.:5..7:..65.5.6.1:.51736.656.5 6:: 360.8.7...9 &0.96066915.5..9....9.3:665 5.9736.96.:765.399::65 09.11.9/..: .73.3::.0699:76515 8.. :0.09.:466.5:.7.67961/...1.97907.07.56.3:6.95 6175150 !64/9 .

:097.65 ...0.65 .59.9.9.51647.1:097.3..51:44.7....9.:06507.65.65..9:65 .65.96507.

:9:53.../..0.9.:.50 5503.::: 551:097..9.9/.3.3::6..:1 0.0.:: :0::65 &44.9.0.30..65 5.91.9:65::0945./.:.9 !64/9 .5 195.93.5/.03.655.::0647...0.9.34..

"510.65: .955.

964.

.65: #6:.955.:0.::47.515...473#.473:53.9.:.14 9505736:67:.51/.

964.

.:16959.655. .47376:...3.

.3.45550 59...:..473:65351. 69:70.65.3.65 #6:.

:1./.61933.

796::.95559.5. 95054.:: !64/9 .3..9.3:65./.3. /.. 59.:: .73/.02.65..73/..45559.65: ..02..6.9/..165/.9/....053..3:65.61:659.

9: .0.647.9:6565.691 9:65&7.0 !64/9 .

.3.:097. 59.51#.333 556 6507.9../:.:.:.:16553.65% &43.4336..0 !64/9 .3.679610.1111.65 50945.0.0.9...9:.516.:.1455.619% 565%% 49065. 59.65%.9736:670..9/.34559:65/.51691:.61.50945...65.3193.:.6.9.5/..593.7731.3 5694.4735 7..655..333.

0.:097.9..1:097..9.59...51:44.65.9:65 ..3.:06507..65.65 .9.65 .96507.51647.65.7..

:.9.9.0.9.03.9..34.65 5.30.93..:1 0..0.3..5 195.::0647.9 !64/9 .:.9/.655./..:9:53.9:65::0945.:: :0::65 &44.0..0..3::6.5/./.50 5503.91.::: 551:097...

.9 6507.9.65 "#.9.511:0945..1:097.0.650.&44.65.

/....:1: .9/.

656" 5..9 1../...65.7:61.510647..3.9.0.9.455606473.34.33345561:097..:.. !64/9 .9:65 551:097.517.0.:9:53.9.65 :097.65 05.47345.695.0.:: :0::65 50945..30.:..3.1510..

.51 .5 .! 9065..9/.%950: .

1510..:: 5 #.65593.../.31.695..:2.65..

31.:: 7./.796...51 9.:.69:5631:0695.&.

19.3 56961.51"#. '#9:: & .51 ...96:5...05636 & "%0691.

.9..51 .65 05.9.43..9/.. .

695...3.5: 5631..51..5595 .159.:: '9.656925631 1:0699643.91./..

063 7..955964.90./3.51.9..0. 33.14.3 5.#9::&44.33507796.69 .473: 5 0.&050.95559. 6/.61:69 3.:.51 :..3:2 0647.3:2.:.05.51% & 0..1 /9#9:: ' ..51&...! 69 #96/.96:30.35.9.3 1..0:695595.

35967. 69.5 %0..& .5.6 #336 .3...9..4.65.4.6959.9..19 6:69.55 9.51 #9.: . . 52..65679.0/93.9.

/ 096::.

./.51:/..

515631:069... 55.6...3: .

51! 9065 . .5 ..

....93:5 93.1951:06968..5.5: 5631.31.65../..:: '9.51..5595.

!64/9 .

7696.5...%950: 065...51 7369. .656.9/.

1510....1 #. 455 5 .6551..:2.695..

&.:.51..41.9:.50:55631:069.51% . 557.69: 1..796# &4..

4:694551:.0.51% ! 369.35..0...&.51 095 7731 3.33 5699.9.50.:...3::911 #95. '#9:: % 65:65.

/.:16..:.33507796.3.3:2.35.69 . .65.955 .65.4/91 ' .61.69.3.0.033 9:65:7.51 6.0..: !692! .3 1.39:53.::.9&30..5 ..140#/3:9: % & 0.95559.6933.05.3:2 . 55 390.514..35.55 ' .616366510.0.7796.063 69.0:0.91.90 9..345..511..65695631:069.4.955 5 0.51..033 59.3350.

05.955 09.9: .33 % $53.65610:65.05.955. ' .5 510.033 .

51 5/.5.0.4.5.# !64/9 ..137..59.65 #3.6557945.69.4 . &/9.

?.34: !64/9 ..7 0: : 0.514/662 %..

Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue listening from where you left off, or restart the preview.

scribd