Data Mining

EL
Week 1: Introduction, Association Rules

PT
Pabitra Mitra

N
Computer Science and Engineering, IIT Kharagpur
Email: pabitra@gmail.com

1
Course Outline:
• Introduction: KDD Process
• Data Preprocessing
• Association Rule Mining

EL
• Classification

PT
• Clustering and Anomaly Detection
• Regression
• Case Studies
N
2
 Data Mining

EL
Introduction

PT
Pabitra Mitra

N
Computer Science and Engineering, IIT Kharagpur

3
 Why Data Mining?
• The Explosive Growth of Data: from terabytes to petabytes
– Data collection and data availability
• Automated data collection tools, database systems, Web, computerized society
– Major sources of abundant data

EL
• Business: Web, e-commerce, transactions, stocks, …

PT
• Science: Remote sensing, bioinformatics, scientific simulation, …
• Society and everyone: news, digital cameras, YouTube

N
• We are drowning in data, but starving for knowledge!
• “Necessity is the mother of invention”—Data mining—Automated analysis of massive data
 What Is Data Mining?
• Data mining (knowledge discovery from data)
– Extraction of interesting (non-trivial, implicit, previously unknown and
potentially useful) patterns or knowledge from huge amount of data

EL
• Alternative names
– Knowledge discovery (mining) in databases (KDD), knowledge extraction,

PT
data/pattern analysis, data archeology, data dredging, information
harvesting, business intelligence, etc.

N
• Watch out: Is everything “data mining”?
– Simple search and query processing
– (Deductive) expert systems

5
 N
PT
6 EL
 Data Mining: Confluence of Multiple Disciplines

Database
Statistics
Technology

EL
Machine Visualization

PT
Data Mining
Learning

N
Pattern
Other
Recognition Algorithm Disciplines

7
 Why Not Traditional Data Analysis?
• Tremendous amount of data
– Algorithms must be highly scalable to handle such as tera-bytes of data
• High-dimensionality of data
– Micro-array may have tens of thousands of dimensions

EL
• High complexity of data
– Data streams and sensor data

PT
– Time-series data, temporal data, sequence data

N
– Structure data, graphs, social networks and multi-linked data
– Heterogeneous databases and legacy databases
– Spatial, spatiotemporal, multimedia, text and Web data

8
 Data Mining: On What Kinds of Data?
• Database-oriented data sets and applications
– Relational database, data warehouse, transactional database
• Advanced data sets and advanced applications
– Data streams and sensor data
– Time-series data, temporal data, sequence data (incl. bio-sequences)

EL
– Structure data, graphs, social networks and multi-linked data

PT
– Object-relational databases
– Heterogeneous databases and legacy databases

N
– Spatial data and spatiotemporal data
– Multimedia database
– Text databases
– The World-Wide Web

9
 Data Mining Functionalities
• Multidimensional concept description: Characterization and discrimination
– Generalize, summarize, and contrast data characteristics, e.g., dry vs. wet
regions
• Frequent patterns, association, correlation vs. causality

EL
– Tea  Sugar [0.5%, 75%] (Correlation or causality?)
• Classification and prediction

PT
– Construct models (functions) that describe and distinguish classes or
concepts for future prediction

N
• E.g., classify countries based on (climate), or classify cars based on (gas
mileage)
– Predict some unknown or missing numerical values

10
 Data Mining Functionalities
• Cluster analysis
– Class label is unknown: Group data to form new classes, e.g., cluster houses to find
distribution patterns
– Maximizing intra-class similarity & minimizing interclass similarity
• Outlier analysis

EL
– Outlier: Data object that does not comply with the general behavior of the data
– Noise or exception? Useful in fraud detection, rare events analysis

PT
• Trend and evolution analysis
– Trend and deviation: e.g., regression analysis

N
– Sequential pattern mining: e.g., digital camera  large SD memory
– Periodicity analysis
– Similarity-based analysis
• Other pattern-directed or statistical analyses

11
 Major Issues in Data Mining
• Mining methodology
– Mining different kinds of knowledge from diverse data types, e.g., bio, stream, Web
– Performance: efficiency, effectiveness, and scalability
– Pattern evaluation: the interestingness problem
– Incorporation of background knowledge
–

EL
Handling noise and incomplete data
– Parallel, distributed and incremental mining methods
– Integration of the discovered knowledge with existing one: knowledge fusion

PT
• User interaction
– Data mining query languages and ad-hoc mining

N
– Expression and visualization of data mining results
– Interactive mining of knowledge at multiple levels of abstraction
• Applications and social impacts
– Domain-specific data mining & invisible data mining
– Protection of data security, integrity, and privacy

12
Architecture: Typical Data Mining System
Graphical User Interface

Pattern Evaluation
Knowledge

EL
Data Mining Engine -Base

PT
Database or Data Warehouse
Server

N
data cleaning, integration, and selection

Database Data World-Wide Other Info

Warehouse Web Repositories

13
 KDD Process: Summary
• Learning the application domain
– relevant prior knowledge and goals of application
• Creating a target data set: data selection
• Data cleaning and preprocessing: (may take 60% of effort!)
• Data reduction and transformation

EL
– Find useful features, dimensionality/variable reduction, invariant representation
• Choosing functions of data mining

PT
– summarization, classification, regression, association, clustering
• Choosing the mining algorithm(s)

N
• Data mining: search for patterns of interest
• Pattern evaluation and knowledge presentation
– visualization, transformation, removing redundant patterns, etc.
• Use of discovered knowledge

14
End of Introduction

EL
PT
N
15
 Data Mining

EL
Data Preprocessing

PT
Pabitra Mitra

N
Computer Science and Engineering, IIT Kharagpur

16
 What is Data?
• Collection of data objects and their Attributes
attributes
• An attribute is a property or Tid Refund Marital Taxable
characteristic of an object Status Income Cheat

EL
– Examples: eye color of a person, 1 Yes Single 125K No

temperature, etc. 2 No Married 100K No

PT
– Attribute is also known as variable, 3 No Single 70K No
4 Yes Married 120K No
field, characteristic, or feature
5 No Divorced 95K Yes
• A collection of attributes describe Objects

N
6 No Married 60K No
an object 7 Yes Divorced 220K No

– Object is also known as record, 8 No Single 85K Yes

point, case, sample, entity, or 9 No Married 75K No

instance 10
10 No Single 90K Yes
 Types of Attributes
• There are different types of attributes
– Nominal
• Examples: ID numbers, eye color, zip codes
– Ordinal

EL
• Examples: rankings (e.g., taste of potato chips on a scale from 1-

PT
10), grades, height in {tall, medium, short}
– Interval

N
• Examples: calendar dates, temperatures in Celsius or Fahrenheit.
– Ratio
• Examples: temperature in Kelvin, length, time, counts
 Properties of Attribute Values
• The type of an attribute depends on which of the following
properties it possesses:
– Distinctness: = 

EL
– Order: < >
– Addition: + -

PT
– Multiplication: */

N
– Nominal attribute: distinctness
– Ordinal attribute: distinctness & order
– Interval attribute: distinctness, order & addition
– Ratio attribute: all 4 properties
Attribute Description Examples Operations
Type
Nominal The values of a nominal attribute are just zip codes, employee mode, entropy,
different names, i.e., nominal attributes ID numbers, eye color, contingency
provide only enough information to
sex: {male, female} correlation, 2 test
distinguish one object from another. (=, )

Ordinal The values of an ordinal attribute hardness of minerals, median, percentiles,

provide enough information to order {good, better, best}, rank correlation,

EL
objects (< >). grades, street numbers run tests, sign tests

PT
Interval For interval attributes, the differences calendar dates, mean, standard
between values are meaningful, i.e., a unit temperature in Celsius deviation, Pearson's
of measurement exists.
or Fahrenheit correlation, t and F

N
(+, - )
tests
Ratio For ratio variables, both differences temperature in Kelvin, geometric mean,
and ratios are meaningful. (*, /) monetary quantities, harmonic mean,
counts, age, mass, length,
percent variation
electrical current
 Discrete and Continuous Attributes
• Discrete Attribute
– Has only a finite or countably infinite set of values
– Examples: zip codes, counts, or the set of words in a collection of
documents
– Often represented as integer variables.

EL
– Note: binary attributes are a special case of discrete attributes

PT
• Continuous Attribute
– Has real numbers as attribute values

N
– Examples: temperature, height, or weight.
– Practically, real values can only be measured and represented using a
finite number of digits.
– Continuous attributes are typically represented as floating-point variables.
 Types of data sets
• Record
– Data Matrix
– Document Data
– Transaction Data
• Graph

EL
– World Wide Web

PT
– Molecular Structures
• Ordered

N
– Spatial Data
– Temporal Data
– Sequential Data
– Genetic Sequence Data
 Record Data
• Data that consists of a collection of records, each of which consists of a
fixed set of attributes Tid Refund Marital Taxable
Status Income Cheat

1 Yes Single 125K No

EL
2 No Married 100K No
3 No Single 70K No

PT
4 Yes Married 120K No
5 No Divorced 95K Yes
6 No Married 60K No

N
7 Yes Divorced 220K No
8 No Single 85K Yes
9 No Married 75K No
10 No Single 90K Yes
10
 Data Matrix
• If data objects have the same fixed set of numeric attributes, then
the data objects can be thought of as points in a multi-dimensional
space, where each dimension represents a distinct attribute

• Such data set can be represented by an m by n matrix, where there

EL
are m rows, one for each object, and n columns, one for each

PT
attribute
Projection Projection Distance Load T hickness

N
of x Load of y load

10.23 5.27 15.22 2.7 1.2

12.65 6.25 16.22 2.2 1.1
 Text Data
• Each document becomes a `term' vector,
– each term is a component (attribute) of the vector,
– the value of each component is the number of times the corresponding term
occurs in the document.

EL

timeout

season
coach

game
PT
score
team

ball

lost
pla

wi
n
N y
Document 1 3 0 5 0 2 6 0 2 0 2

Document 2 0 7 0 2 1 0 0 3 0 0

Document 3 0 1 0 0 1 2 2 0 3 0
 Transaction Data
• A special type of record data, where
– each record (transaction) involves a set of items.
– For example, consider a grocery store. The set of products purchased

EL
by a customer during one shopping trip constitute a transaction, while
the individual products that were purchased are the items.

PT
TID Items
1 Bread, Coke, Milk

N
2 Beer, Bread
3 Beer, Coke, Diaper, Milk
4 Beer, Bread, Diaper, Milk
5 Coke, Diaper, Milk
 Graph Data
• Examples: Facebook graph and HTML Links

EL
2
5 1

PT
2

N
5
 Ordered Data
• Genomic sequence data
GGTTCCGCCTTCAGCCCCGCGCC

EL
CGCAGGGCCCGCCCCGCGCCGTC
GAGAAGGGCCCGCCTGGCGGGCG

PT
GGGGGAGGCGGGGCCGCCCGAGC
CCAACCGAGTCCGACCAGGTGCC
CCCTCTGCTCGGCCTAGACCTGA

N
GCTCATTAGGCGGCAGCGGACAG
GCCAAGTAGAACACGCGAAGCGC
TGGGCTGCCTGCTGCGACCAGGG
 Data Quality
• What kinds of data quality problems?
• How can we detect problems with the data?

EL
• What can we do about these problems?

PT
• Examples of data quality problems:
– Noise and outliers
– missing values
– duplicate data
N
 Noise
• Noise refers to modification of original values
– Examples: distortion of a person’s voice when talking on a poor phone
and “snow” on television screen

EL
PT
N Two Sine Waves + Noise
Two Sine Waves
 Outliers
• Outliers are data objects with characteristics that are
considerably different than most of the other data objects in

EL
the data set

PT
N
 Missing Values
• Reasons for missing values
– Information is not collected
(e.g., people decline to give their age and weight)
– Attributes may not be applicable to all cases

EL
(e.g., annual income is not applicable to children)

PT
• Handling missing values
– Eliminate Data Objects

N
– Estimate Missing Values
– Ignore the Missing Value During Analysis
– Replace with all possible values (weighted by their probabilities)
 Duplicate Data
• Data set may include data objects that are duplicates, or
almost duplicates of one another
– Major issue when merging data from heterogenous sources

EL
• Examples:

PT
– Same person with multiple email addresses

• Data cleaning
N
– Process of dealing with duplicate data issues
 Data Preprocessing
• Aggregation
• Sampling

EL
• Dimensionality Reduction
•

PT
Feature subset selection
• Feature creation
•
•
N
Discretization and Binarization
Attribute Transformation
 Aggregation
• Combining two or more attributes (or objects) into a
single attribute (or object)
• Purpose

EL
– Data reduction
• Reduce the number of attributes or objects

PT
– Change of scale

N
• Cities aggregated into regions, states, countries, etc
– More “stable” data
• Aggregated data tends to have less variability
 Sampling
• Sampling is the main technique employed for data selection.
– It is often used for both the preliminary investigation of the data and the final
data analysis.

EL
• Statisticians sample because obtaining the entire set of data of interest

PT
is too expensive or time consuming.

N
• Sampling is used in data mining because processing the entire set of
data of interest is too expensive or time consuming.
 Sample Size

EL
PT
N
8000 points 2000 Points 500 Points
 Sampling …
• The key principle for effective sampling is the
following:

EL
– using a sample will work almost as well as using the

PT
entire data sets, if the sample is representative
– A sample is representative if it has approximately the
N
same property (of interest) as the original set of data
 Types of Sampling
• Simple Random Sampling
– There is an equal probability of selecting any particular item

• Sampling without replacement

EL
– As each item is selected, it is removed from the population

PT
• Sampling with replacement
– Objects are not removed from the population as they are selected for the sample.
•

N
In sampling with replacement, the same object can be picked up more than once

• Stratified sampling
– Split the data into several partitions; then draw random samples from each
partition
 Curse of Dimensionality
• When dimensionality increases, data becomes increasingly sparse
in the space that it occupies

EL
• Definitions of density and distance between points, which is

PT
critical for clustering and outlier detection, become less
meaningful

N
 Dimensionality Reduction
• Purpose:
– Avoid curse of dimensionality
– Reduce amount of time and memory required by data mining
algorithms

EL
– Allow data to be more easily visualized
– May help to eliminate irrelevant features or reduce noise

PT
• Techniques

N
– Principle Component Analysis
– Singular Value Decomposition
– Others: supervised and non-linear techniques
 Discretization

EL
Data Equal interval width

PT
N
Equal frequency K-means
 Attribute Transformation
• A function that maps the entire set of values of
a given attribute to a new set of replacement

EL
values such that each old value can be

PT
identified with one of the new values

N
– Simple functions: xk, log(x), ex, |x|
– Standardization and Normalization
 Similarity and Dissimilarity
• Similarity
– Numerical measure of how alike two data objects are.
– Is higher when objects are more alike.

EL
– Often falls in the range [0,1]
• Dissimilarity

PT
– Numerical measure of how different are two data objects
– Lower when objects are more alike

N
– Minimum dissimilarity is often 0
– Upper limit varies
• Proximity refers to a similarity or dissimilarity
 Similarity/Dissimilarity for Simple Attributes
p and q are the attribute values for two data objects.

EL
PT
N
 Euclidean Distance
• Euclidean Distance
n
dist   ( pk  qk )2
k 1

EL
PT
Where n is the number of dimensions (attributes) and pk and qk are,
respectively, the kth attributes (components) or data objects p and q.

N
• Standardization is necessary, if scales differ.
 Mahalanobis Distance
mahalanobis( p, q)  ( p  q) 1 ( p  q)T

 is the covariance matrix of the

input data X

EL
n
1
 j ,k   ( X ij  X j )( X ik  X k )
n  1 i 1

PT
N
For red points, the Euclidean distance is 14.7, Mahalanobis distance is 6.
 Cosine Similarity
• If d1 and d2 are two document vectors, then
cos( d1, d2 ) = (d1  d2) / ||d1|| ||d2|| ,
where  indicates vector dot product and || d || is the length of vector d.
• Example:

EL
d1 = 3 2 0 5 0 0 0 2 0 0
d2 = 1 0 0 0 0 0 0 1 0 2

PT
d1  d2= 3*1 + 2*0 + 0*0 + 5*0 + 0*0 + 0*0 + 0*0 + 2*1 + 0*0 + 0*2 = 5

N
||d1|| = (3*3+2*2+0*0+5*5+0*0+0*0+0*0+2*2+0*0+0*0)0.5 = (42) 0.5 = 6.481
||d2|| = (1*1+0*0+0*0+0*0+0*0+0*0+0*0+1*1+0*0+2*2) 0.5 = (6) 0.5 = 2.245

cos( d1, d2 ) = .3150

 Similarity Between Binary Vectors
• Common situation is that objects, p and q, have only
binary attributes
• Compute similarities using the following quantities
M01 = the number of attributes where p was 0 and q was 1
M10 = the number of attributes where p was 1 and q was 0

EL
M00 = the number of attributes where p was 0 and q was 0
M11 = the number of attributes where p was 1 and q was 1

PT
• Simple Matching and Jaccard Coefficients
SMC = number of matches / number of attributes

N
= (M11 + M00) / (M01 + M10 + M11 + M00)

J = number of 11 matches / number of not-both-zero attributes values

= (M11) / (M01 + M10 + M11)
 Correlation
• Correlation measures the linear relationship between objects
• To compute correlation, we standardize data objects, p and q, and then
take their dot product

EL
  ( pk  mean( p)) / std( p)
pk

PT
  (qk  mean(q)) / std(q)
qk

N
correlation( p, q)  p  q
Visually Evaluating Correlation

EL
Scatter plots

PT
showing the
similarity from –1

N
to 1.
 EL
End of Data Preprocessing

PT
N
 Data Mining

EL
Association Rules

PT
Pabitra Mitra

N
Computer Science and Engineering

53
 Association Rule Mining
• Given a set of transactions, find rules that will predict the
occurrence of an item based on the occurrences of other items in
the transaction

EL
Market-Basket transactions
Example of Association Rules

PT
TID Items
1 Bread, Milk {Diaper}  {Beer},
{Milk, Bread}  {Eggs,Coke},

N
2 Bread, Diaper, Beer, Eggs
3 Milk, Diaper, Beer, Coke {Beer, Bread}  {Milk},
4 Bread, Milk, Diaper, Beer Implication means co-occurrence,
5 Bread, Milk, Diaper, Coke not causality!
 Definition: Frequent Itemset
• Itemset
– A collection of one or more items
• Example: {Milk, Bread, Diaper}
– k-itemset TID Items

• An itemset that contains k items 1 Bread, Milk

EL
• Support count () 2 Bread, Diaper, Beer, Eggs
– Frequency of occurrence of an itemset 3 Milk, Diaper, Beer, Coke

PT
– E.g. ({Milk, Bread,Diaper}) = 2 4 Bread, Milk, Diaper, Beer
• Support 5 Bread, Milk, Diaper, Coke

N
– Fraction of transactions that contain an itemset
– E.g. s({Milk, Bread, Diaper}) = 2/5
• Frequent Itemset
– An itemset whose support is greater than or equal to a
minsup threshold
 TID Items
1 Bread, Milk
 Association Rule
2 Bread, Diaper, Beer, Eggs
– An implication expression of the form X  Y, where X
3 Milk, Diaper, Beer, Coke
and Y are itemsets
4 Bread, Milk, Diaper, Beer
– Example:
5 Bread, Milk, Diaper, Coke
{Milk, Diaper}  {Beer}

EL
 Rule Evaluation Metrics
– Support (s) Example:

PT
 Fraction of transactions that contain both X and {Milk, Diaper }  Beer
Y
 ( Milk , Diaper, Beer )

N
2
– Confidence (c) s   0.4
 Measures how often items in Y
|T| 5
appear in transactions that  ( Milk, Diaper, Beer ) 2
contain X c   0.67
 ( Milk , Diaper ) 3
 Association Rule Mining Task
• Given a set of transactions T, the goal of association rule mining is
to find all rules having
– support ≥ minsup threshold

EL
– confidence ≥ minconf threshold

PT
• Brute-force approach:
– List all possible association rules

N
– Compute the support and confidence for each rule
– Prune rules that fail the minsup and minconf thresholds
 Computationally prohibitive!
 Mining Association Rules
• Two-step approach:
1. Frequent Itemset Generation
– Generate all itemsets whose support  minsup

EL
2. Rule Generation

PT
– Generate high confidence rules from each frequent
itemset, where each rule is a binary partitioning of a

N
frequent itemset
• Frequent itemset generation is still
computationally expensive
 Frequent Itemset Generation
null

A B C D E

EL
AB AC AD AE BC BD BE CD CE DE

PT
ABC ABD ABE ACD ACE ADE BCD BCE BDE CDE

ABCD ABCE N ABDE ACDE BCDE

Given d items, there are
2d possible candidate
itemsets
ABCDE
 Frequent Itemset Generation
• Brute-force approach:
– Each itemset in the lattice is a candidate frequent itemset
– Count the support of each candidate by scanning the database
Transactions List of

EL
Candidates
TID Items
1 Bread, Milk

PT
2 Bread, Diaper, Beer, Eggs
N 3 Milk, Diaper, Beer, Coke M
4 Bread, Milk, Diaper, Beer

N
5 Bread, Milk, Diaper, Coke
w
– Match each transaction against every candidate
– Complexity ~ O(NMw) => Expensive since M = 2d !!!
 Frequent Itemset Generation Strategies
• Reduce the number of candidates (M)
– Complete search: M=2d
– Use pruning techniques to reduce M
• Reduce the number of transactions (N)

EL
– Reduce size of N as the size of itemset increases
– Used by DHP and vertical-based mining algorithms

PT
N
• Reduce the number of comparisons (NM)
– Use efficient data structures to store the candidates or transactions
– No need to match every candidate against every transaction
 Reducing Number of Candidates
• Apriori principle:
– If an itemset is frequent, then all of its subsets must also be
frequent

EL
• Apriori principle holds due to the following property of the
support measure:

PT
X , Y : ( X  Y )  s( X )  s(Y )

N
– Support of an itemset never exceeds the support of its subsets
– This is known as the anti-monotone property of support
 Illustrating Apriori Principle
null

A B C D E

EL
AB AC AD AE BC BD BE CD CE DE

PT
Found to be
Infrequent ABC ABD ABE ACD ACE ADE BCD BCE BDE CDE

Pruned
supersets
N ABCD ABCE ABDE ACDE BCDE

ABCDE
 Illustrating Apriori Principle
Item Count Items (1-itemsets)
Bread 4
Coke 2
Milk
Beer
4
3
Itemset Count Pairs (2-itemsets)
{Bread,Milk} 3
Diaper 4

EL
{Bread,Beer} 2
Eggs 1
{Bread,Diaper} 3 (No need to generate
{Milk,Beer} 2
candidates involving Coke

PT
{Milk,Diaper} 3
{Beer,Diaper} 3 or Eggs)
Minimum Support = 3
Triplets (3-itemsets)

N
If every subset is considered, Itemset
{Bread,Milk,Diaper}
Count
3
6C + 6C + 6C = 41
1 2 3
With support-based pruning,
6 + 6 + 1 = 13
 Apriori Algorithm
• Method:
– Let k=1
– Generate frequent itemsets of length 1

EL
– Repeat until no new frequent itemsets are identified
• Generate length (k+1) candidate itemsets from length k frequent

PT
itemsets
• Prune candidate itemsets containing subsets of length k that are

N
infrequent
• Count the support of each candidate by scanning the DB
• Eliminate candidates that are infrequent, leaving only those that
are frequent
 Factors Affecting Complexity
• Choice of minimum support threshold
– lowering support threshold results in more frequent itemsets
– this may increase number of candidates and max length of frequent itemsets
• Dimensionality (number of items) of the data set
– more space is needed to store support count of each item

EL
– if number of frequent items also increases, both computation and I/O costs may
also increase

PT
• Size of database
– Apriori makes multiple passes, run time of algorithm increase with number of

N
transactions
• Average transaction width
– This may increase max length of frequent itemsets and traversals of hash tree
(number of subsets in a transaction increases with its width)
 Rule Generation
• How to efficiently generate rules from frequent itemsets?
– In general, confidence does not have an anti-monotone property
c(ABC D) can be larger or smaller than c(AB D)

EL
– But confidence of rules generated from the same itemset has an anti-
monotone property

PT
– e.g., L = {A,B,C,D}:
c(ABC  D)  c(AB  CD)  c(A  BCD)

N
• Confidence is anti-monotone w.r.t. number of items on the RHS of the rule
 Rule Generation for Apriori Algorithm
Lattice of rules ABCD=>{ }
Low
Confidence
Rule BCD=>A ACD=>B ABD=>C ABC=>D

EL
PT
CD=>AB BD=>AC BC=>AD AD=>BC AC=>BD AB=>CD

Pruned
Rules D=>ABC
N
C=>ABD B=>ACD A=>BCD
 Rule Generation for Apriori Algorithm
• Candidate rule is generated by merging two rules that share the same prefix
in the rule consequent
CD=>AB BD=>AC

EL
• join(CD=>AB,BD=>AC)
would produce the candidate

PT
rule D => ABC

N
• Prune rule D=>ABC if its
D=>ABC
subset AD=>BC does not have
high confidence
 Pattern Evaluation
• Association rule algorithms tend to produce too many rules
– many of them are uninteresting or redundant
– Redundant if {A,B,C}  {D} and {A,B}  {D}

EL
have same support & confidence

PT
• Interestingness measures can be used to prune/rank the derived
patterns

N
• In the original formulation of association rules, support &
confidence are the only measures used
Application of Interestingness Measure
Interestingness
Measures

EL
PT
N
 Computing Interestingness Measure
• Given a rule X  Y, information needed to compute rule interestingness
can be obtained from a contingency table
Contingency table for supports X  Y

EL
Y Y
X f11 f10 f1+

PT
X f01 f00 fo+
f+1 f+0 |T|

N
Used to define various measures
 support, confidence, lift, Gini,
J-measure, etc.
 Statistical Independence
• Population of 1000 students
– 600 students know how to swim (S)
– 700 students know how to bike (B)

EL
– 420 students know how to swim and bike (S,B)

PT
– P(SB) = 420/1000 = 0.42
– P(S)  P(B) = 0.6  0.7 = 0.42

N
– P(SB) = P(S)  P(B) => Statistical independence
– P(SB) > P(S)  P(B) => Positively correlated
– P(SB) < P(S)  P(B) => Negatively correlated
 Statistical-based Measures
• take into account statistical dependence
P (Y | X )
Lift 
P (Y )

EL
P( X , Y )
Interest 

PT
P ( X ) P (Y )
PS  P ( X , Y )  P ( X ) P (Y )

  coefficient  N P ( X , Y )  P ( X ) P (Y )
P ( X )[1  P ( X )]P (Y )[1  P (Y )]
 Example: Lift/Interest
Coffee Coffee
Tea 15 5 20
Tea 75 5 80

EL
90 10 100

PT
Association Rule: Tea  Coffee
Confidence= P(Coffee|Tea) = 0.75

N
but P(Coffee) = 0.9
 Lift = 0.75/0.9= 0.8333 (< 1, therefore is negatively associated)
There are lots of
measures proposed in
the literature

Some measures are

good for certain

EL
applications, but not for
others

PT
What criteria should we

N
use to determine
whether a measure is
good or bad?

What about Apriori-

style support based
Subjective Interestingness Measure
• Objective measure:
– Rank patterns based on statistics computed from data
– e.g., 21 measures of association (support, confidence, Laplace, Gini,

EL
mutual information, Jaccard, etc).

PT
• Subjective measure:
– Rank patterns according to user’s interpretation

N
• A pattern is subjectively interesting if it contradicts the
expectation of a user (Silberschatz & Tuzhilin)
• A pattern is subjectively interesting if it is actionable
(Silberschatz & Tuzhilin)
 Interestingness via Unexpectedness
• Need to model expectation of users (domain knowledge)

+ Pattern expected to be frequent

- Pattern expected to be infrequent

EL
Pattern found to be frequent
Pattern found to be infrequent

PT
+ - Expected Patterns

N
- + Unexpected Patterns

• Need to combine expectation of users with evidence from data (i.e., extracted patterns)
 EL
End of Association Rule

PT
N