Unit III: Concept Description:

Characterization and Comparison

 Unit III: Concept Description:
Characterization and Comparison
 DMDL Primitives and Queries
 Architectures of DM
 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
 Summary
 What is Concept Description?

 Descriptive vs. predictive data mining

– Descriptive mining: describes concepts or task-relevant data
sets in concise, summative, 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
 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. 1
2
3
4
Conceptual levels
5

– Approaches:
 Data cube approach(OLAP approach)


 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(AOI), but
also Attribute Relevance Analysis (ARA), dicing, slicing, roll-
up and drill-down
– Cost of cube computation and the nontrivial storage overhead
 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.
 Characterization: Data Cube Approach
(without using Attribute Oriented-
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
 Attribute-Oriented
Induction
 Proposed in 1989 (KDD ‘89 workshop)
 Not confined to categorical data nor particular
measures.
 How it is done?
1. Collect the task-relevant data( initial relation) using a
relational database query
2. Perform generalization by attribute removal or
attribute generalization.
3. Apply aggregation by merging identical, generalized
tuples and accumulating their respective counts.
4. 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.
1
 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.
1
 Class Characterization: An Example
Name Gender Major Birth-Place Birth_date Residence Phone # GPA

Initial Jim M CS Vancouver,BC, 8-12-76 3511 Main St., 687-4598 3.67

Woodman Canada Richmond
Relation Scott M CS Montreal, Que, 28-7-75 345 1st Ave., 253-9106 3.70
Lachance Canada Richmond
Laura Lee F Physics Seattle, WA, USA 25-8-70 125 Austin Ave., 420-5232 3.83
… … … … … Burnaby … …
…
Removed Retained Sci,Eng, Country Age range City Removed Excl,
Bus VG,..
Gender Major Birth_region Age_range Residence GPA Count
Prime M Science Canada 20-25 Richmond Very-good 16
Generalized F Science Foreign 25-30 Burnaby Excellent 22
Relation … … … … … … …

Birth_Region
Canada Foreign Total
Gender
M 16 14 30
F 10 22 32
Total 26 36 62

1
 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” }

1
 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.,
grad( x)  male( x) 
birth_ region( x) "Canada"[t :53%] birth_ region( x) " foreign"[t : 47%].
1
 Presentation of Generalized Results
(continued)
 t-weight:
– Interesting measure that describes the typicality of
 each disjunct in the rule
 each tuple in the corresponding generalized relation
n

t_weight = count(qa)/ å count(qi)

i =1

 n – number of tuples for target class for generalized

relation
 qi … qn – tuples for target class in generalized relation
 qa is in qi … qn
1
Presentation—Generalized Relation

1
Presentation—Crosstab

1
 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

1
 Attribute relevance analysis (cont’d)

 How?
1. Data Collection
2. Analytical Generalization
 Use information gain analysis (e.g., entropy or other
measures) to identify highly relevant dimensions and
levels.
3. Relevance Analysis
 Sort and select the most relevant dimensions and levels.
4. Attribute-oriented Induction for class description
 On selected dimension/level
5. OLAP operations (e.g. drilling, slicing) on
relevance rules

1
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
– (Chi-Square) 2 contingency table statistics
– uncertainty coefficient

2
 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

2
Training Examples

ICS320
2
 Top-Down Induction of Decision
Tree
Attributes = {Outlook, Temperature, Humidity, Wind}
PlayTennis = {yes, no}

Outlook
sunny overcast rain

Humidity yes Wind

high normal strong weak

no yes no yes

2
 Entropy and Information Gain

 S contains si tuples of class Ci for i = {1, …, m}

 Information measures info required to classify any
m
arbitrary tuple si si
I( s1,s2,...,sm)   log 2
i 1 s s
 Entropy (weighted average) of attribute A with values
{a1,a2,…,av}
v
s1 j  ...  smj
E(A)   I ( s1 j ,..., smj )
j 1 s
 Information gained by branching on attribute A
Gain(A)  I(s 1, s 2 ,..., sm)  E(A)
– >info gained > discriminating attribute

2
 Class Characterization: An Example
Name Gender Major Birth-Place Birth_date Residence Phone # GPA

Initial Jim M CS Vancouver,BC, 8-12-76 3511 Main St., 687-4598 3.67

Woodman Canada Richmond
Relation Scott M CS Montreal, Que, 28-7-75 345 1st Ave., 253-9106 3.70
Lachance Canada Richmond
Laura Lee F Physics Seattle, WA, USA 25-8-70 125 Austin Ave., 420-5232 3.83
… … … … … Burnaby … …
…
Removed Retained Sci,Eng, Country Age range City Removed Excl,
Bus VG,..
Gender Major Birth_region Age_range Residence GPA Count
Prime M Science Canada 20-25 Richmond Very-good 16
Generalized F Science Foreign 25-30 Burnaby Excellent 22
Relation … … … … … … …

Birth_Region
Canada Foreign Total
Gender
M 16 14 30
F 10 22 32
Total 26 36 62

2
 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(ai) = concept hierarchies on ai
– Ui = attribute analytical thresholds for ai
– Ti = attribute generalization thresholds for ai
– R = attribute relevance threshold
2
 Example: Analytical
Characterization (cont’d)
 1. Data collection
– target class: graduate student
– contrasting class: undergraduate student
 2. Analytical generalization using Ui
– 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

2
 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 ( =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 ( =130)

2
 Example: Analytical characterization
(3)
 3. Relevance analysis
– Calculate expected info required to classify an arbitrary tuple

120 120 130 130

I(s 1, s 2 )  I( 120 ,130 )   log 2  log 2  0.9988
250 250 250 250
– Calculate entropy of each attribute: e.g. major
For major=”Science”: S11=84 S21=42 I(s11,s21)=0.9183
For major=”Engineering”: S12=36 S22=46 I(s12,s22)=0.9892
For major=”Business”: S13=0 S23=42 I(s13,s23)=0

Number of grad Number of undergrad

students in “Science” students in “Science”

84  84  42  42 
I ( s11 , s 21 )   log 2    log 2   0.9183
126  126  126  126 
2
 Example: Analytical Characterization (4)

 Calculate expected info required to classify a given

sample if S is partitioned according to the attribute
0.9183
126 82 0.9892 42 0
E(major)  I ( s11 , s 21 )  I ( s12 , s 22 )  I ( s13, s 23 )  0.7873
250 250 250
 Calculate information gain for each attribute
0.9988 0.7873
Gain(major)  I(s 1, s 2 )  E(major)  0.2115
– Information gain for all attributes
Gain(gender) = 0.0003
Gain(birth_country) = 0.0407
Gain(major) = 0.2115
Gain(gpa) = 0.4490
Gain(age_range) = 0.5971
3
 Example: Analytical characterization
(5)
 4. Initial working relation (W0) derivation
– R (attribute relevance threshold) = 0.1
– remove irrelevant/weakly relevant attributes from candidate
relation => drop gender, birth_country
– remove contrasting class candidate relation
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 W0: Graduate students

 5. Perform attribute-oriented induction on W using T 3

 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.

3
 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

3
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

3
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

3
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

3
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%

3
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: [0.0, 1.0] or [0%, 100%]
count(q a  Cj )
d  weight  m

 count(q
i 1
a  Ci )

3
 Example: Quantitative Description
Rule
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_ 200 20% 100% 800 80% 100% 1000 100% 100%
regions

Crosstab showing associated t-weight, 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, Europe(X) 
(item(X) " TV" ) [t : 25%, d : 40%]  (item(X) " computer" ) [t : 75%, d : 30%]

3
Quantitative Discriminant Rules
 High d-weight in target class indicates that concept
represented by generalized tuple is primarily derived
from target class
 Low d-weight implies concept is derived from
contrasting class
 Threshold can be set to control the display of
interesting tuples
 quantitative discriminant rule form
 X, target_cla ss(X)  condition( X) [d : d_weight]

Read: if X satisfies condition, there is a probability (d-weight)

that x is in the target class

4
 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
4
 Measuring the Central Tendency
1 n
 Mean x   xi n

n i 1 w x i i
– Weighted arithmetic mean x  i 1
n

w i
 Median: A holistic measure i 1

– Middle value if odd number of values, or average of the

middle two values otherwise n / 2  ( f )l
median  L1  ( )c
– estimated by interpolation f median
 Mode
– Value that occurs most frequently in the data
– Unimodal, bimodal, trimodal
– Empirical formula: mean  mode  3  (mean  median)
4
 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)
1 n 1 n
1 n
  
2
s 
2
( xi  x ) 2
 [ xi  ( xi ) 2
]
n  1 i 1 n  1 i 1 n i 1
– Standard deviation s is the square root of variance s2 4
 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 IQR
(interquartile range)
– The median is marked by a line within the box
– Whiskers: two lines outside the box extend to
Minimum and Maximum

4
A Boxplot

4
Mining Descriptive Statistical Measures in
Large Databases

 Variance
s   ( xi  x ) 2   xi2    xi  
2 1 n 1 1 2

n i 1 n 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

4
 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

4
 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

4
 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

4
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

5
 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

5
 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

5
 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.
5