1

CSE300

Data mining and its application and Data mining and its application and usage in medicineusage in medicine

By RadhikaBy Radhika

2

CSE300

Data Mining and MedicineData Mining and Medicine HistoryHistory

Past 20 years with relational databases More dimensions to database queries

earliest and most successful area of data mining Mid 1800s in London hit by infectious disease

Two theories– Miasma theory Bad air propagated disease– Germ theory Water-borne

Advantages– Discover trends even when we don’t understand reasons– Discover irrelevant patterns that confuse than enlighten– Protection against unaided human inference of patterns provide

quantifiable measures and aid human judgment Data Mining

Patterns persistent and meaningful Knowledge Discovery of Data

3

CSE300

The future of data miningThe future of data mining 10 biggest killers in the US10 biggest killers in the US

Data mining = Process of discovery of interesting, Data mining = Process of discovery of interesting, meaningful and actionable patterns hidden in large meaningful and actionable patterns hidden in large amounts of dataamounts of data

4

CSE300

Major Issues in Medical Data MiningMajor Issues in Medical Data Mining Heterogeneity of medical dataHeterogeneity of medical data

Volume and complexity Physician’s interpretation Poor mathematical categorization Canonical Form Solution: Standard vocabularies, interfaces

between different sources of data integrations, design of electronic patient records

Ethical, Legal and Social IssuesEthical, Legal and Social Issues Data Ownership Lawsuits Privacy and Security of Human Data Expected benefits Administrative Issues

5

CSE300

Why Data Preprocessing?Why Data Preprocessing? Patient records consist of clinical, lab parameters, Patient records consist of clinical, lab parameters,

results of particular investigations, specific to tasksresults of particular investigations, specific to tasks Incomplete: lacking attribute values, lacking

certain attributes of interest, or containing only aggregate data

Noisy: containing errors or outliers Inconsistent: containing discrepancies in codes or

names Temporal chronic diseases parameters

No quality data, no quality mining results!No quality data, no quality mining results! Data warehouse needs consistent integration of

quality data Medical Domain, to handle incomplete,

inconsistent or noisy data, need people with domain knowledge

6

CSE300

What is Data Mining? The KDD ProcessWhat is Data Mining? The KDD Process

Data Cleaning

Data Integration

Databases

Data Warehouse

Task-relevant Data

Selection

Data Mining

Pattern Evaluation

7

CSE300

From Tables and Spreadsheets to Data CubesFrom Tables and Spreadsheets to Data Cubes A data warehouse is based on a A data warehouse is based on a multidimensionalmultidimensional data data

model that views data in the form of a model that views data in the form of a data cubedata cube A data cube, such as sales, allows data to be modeled A data cube, such as sales, allows data to be modeled

and viewed in multiple dimensionsand viewed in multiple dimensions Dimension tables, such as item (item_name, brand,

type), or time(day, week, month, quarter, year) Fact table contains measures (such as

dollars_sold) and keys to each of related dimension tables

W. H. Inmon:“A data warehouse is a W. H. Inmon:“A data warehouse is a subject-orientedsubject-oriented, , integratedintegrated, , time-varianttime-variant, and , and nonvolatilenonvolatile collection of collection of data in support of management’s decision-making data in support of management’s decision-making process.”process.”

8

CSE300

Data Warehouse vs. Heterogeneous DBMSData Warehouse vs. Heterogeneous DBMS Data warehouse: update-driven, high performanceData warehouse: update-driven, high performance

Information from heterogeneous sources is integrated in advance and stored in warehouses for direct query and analysis

Do not contain most current information Query processing does not interfere with

processing at local sources Store and integrate historical information Support complex multidimensional queries

9

CSE300

Data Warehouse vs. Operational DBMSData Warehouse vs. Operational DBMS OLTP (on-line transaction processing)OLTP (on-line transaction processing)

Major task of traditional relational DBMS Day-to-day operations: purchasing, inventory,

banking, manufacturing, payroll, registration, accounting, etc.

OLAP (on-line analytical processing)OLAP (on-line analytical processing) Major task of data warehouse system Data analysis and decision making

Distinct features (OLTP vs. OLAP):Distinct features (OLTP vs. OLAP): User and system orientation: customer vs. market Data contents: current, detailed vs. historical,

consolidated Database design: ER + application vs. star + subject View: current, local vs. evolutionary, integrated Access patterns: update vs. read-only but complex

queries

10

CSE300

11

CSE300

Why Separate Data Warehouse?Why Separate Data Warehouse? High performance for both systemsHigh performance for both systems

DBMS tuned for OLTP: access methods, indexing, concurrency control, recovery

Warehouse tuned for OLAP: complex OLAP queries, multidimensional view, consolidation

Different functions and different data:Different functions and different data: Missing data: Decision support requires historical

data which operational DBs do not typically maintain

Data consolidation: DS requires consolidation (aggregation, summarization) of data from heterogeneous sources

Data quality: different sources typically use inconsistent data representations, codes and formats which have to be reconciled

12

CSE300

13

CSE300

14

CSE300

Typical OLAP OperationsTypical OLAP Operations Roll up (drill-up): summarize dataRoll up (drill-up): summarize data

by climbing up hierarchy or by dimension reduction Drill down (roll down): reverse of roll-upDrill down (roll down): reverse of roll-up

from higher level summary to lower level summary or detailed data, or introducing new dimensions

Slice and dice: Slice and dice: project and select

Pivot (rotate): Pivot (rotate): reorient the cube, visualization, 3D to series of 2D planes.

Other operationsOther operations drill across: involving (across) more than one fact table drill through: through the bottom level of the cube to its

back-end relational tables (using SQL)

15

CSE300

16

CSE300

17

CSE300

Multi-Tiered ArchitectureMulti-Tiered Architecture

DataWarehouse

ExtractTransformLoadRefresh

OLAP Engine

AnalysisQueryReportsData mining

Monitor&

IntegratorMetadata

Data Sources Front-End Tools

Serve

Data Marts

Operational DBs

other

sources

Data Storage

OLAP Server

18

CSE300

Steps of a KDD Process Steps of a KDD Process Learning the application domain:Learning the application domain:

relevant prior knowledge and goals of application Creating a target data set: data selectionCreating a target data set: data selection Data cleaning and preprocessing: (may take 60% of effort!)Data cleaning and preprocessing: (may take 60% of effort!) Data reduction and transformation:Data reduction and transformation:

Find useful features, dimensionality/variable reduction, invariant representation.

Choosing functions of data mining Choosing functions of data mining summarization, classification, regression, association,

clustering. Choosing the mining algorithm(s)Choosing the mining algorithm(s) Data mining: search for patterns of interestData mining: search for patterns of interest Pattern evaluation and knowledge presentationPattern evaluation and knowledge presentation

visualization, transformation, removing redundant patterns, etc.

Use of discovered knowledgeUse of discovered knowledge

19

CSE300

Common Techniques in Data MiningCommon Techniques in Data Mining Predictive Data MiningPredictive Data Mining

Most important Classification: Relate one set of variables in data to

response variables Regression: estimate some continuous value

Descriptive Data MiningDescriptive Data Mining Clustering: Discovering groups of similar instances Association rule extraction

Variables/Observations Summarization of group descriptions

20

CSE300

LeukemiaLeukemia Different types of cells look very similarDifferent types of cells look very similar Given a number of samples (patients) Given a number of samples (patients)

can we diagnose the disease accurately? Predict the outcome of treatment? Recommend best treatment based of previous

treatments? Solution: Data mining on micro-array dataSolution: Data mining on micro-array data 38 training patients, 34 testing patients ~ 7000 patient 38 training patients, 34 testing patients ~ 7000 patient

attributesattributes 2 classes: Acute Lymphoblastic Leukemia(ALL) vs 2 classes: Acute Lymphoblastic Leukemia(ALL) vs

Acute Myeloid Leukemia (AML) Acute Myeloid Leukemia (AML)

21

CSE300

Clustering/Instance Based LearningClustering/Instance Based Learning Uses specific instances to perform classification than general Uses specific instances to perform classification than general

IF THEN rulesIF THEN rules Nearest Neighbor classifierNearest Neighbor classifier Most studied algorithms for medical purposesMost studied algorithms for medical purposes Clustering– Partitioning a data set into several groups Clustering– Partitioning a data set into several groups

(clusters) such that(clusters) such that Homogeneity: Objects belonging to the same cluster are

similar to each other Separation: Objects belonging to different clusters are

dissimilar to each other.  Three elements Three elements

The set of objects The set of attributes Distance measure

22

CSE300

Measure the Dissimilarity of ObjectsMeasure the Dissimilarity of Objects

Find best matching instanceFind best matching instance Distance functionDistance function

Measure the dissimilarity between a pair of data objects

Things to considerThings to consider Usually very different for interval-scaled,

boolean, nominal, ordinal and ratio-scaled variables

Weights should be associated with different variables based on applications and data semantic

Quality of a clustering result depends on both the Quality of a clustering result depends on both the distance measuredistance measure adopted and its implementation adopted and its implementation

23

CSE300

Minkowski DistanceMinkowski Distance Minkowski distance: a generalizationMinkowski distance: a generalization

If q = 2, d is Euclidean distanceIf q = 2, d is Euclidean distance If q = 1, d is Manhattan distanceIf q = 1, d is Manhattan distance

)0(||...||||),(2211

qqj

xi

xj

xi

xj

xi

xjid q

pp

qq

xi

xj

q=2 q=16

6

128.48

Xi (1,7)

Xj(7,1)

24

CSE300

Binary VariablesBinary Variables A contingency table for binary dataA contingency table for binary data

Simple matching coefficientSimple matching coefficient

dcbacb jid

),(

pdbcasum

dcdc

baba

sum

0

1

01

Object i

Object j

25

CSE300

Dissimilarity between Binary VariablesDissimilarity between Binary Variables ExampleExample

A1A1 A2A2 A3A3 A4A4 A5A5 A6A6 A7A7

Object 1Object 1 11 00 11 11 11 00 00

Object 2Object 2 11 11 11 00 00 00 11

Object 1

Object 2

11 00 sumsum

11 22 22 44

00 22 11 33

sumsum 44 33 77 7

41222

22)2

,1

(

OOd

26

CSE300

K-nearest neighbors algorithmK-nearest neighbors algorithm InitializationInitialization

Arbitrarily choose k objects as the initial cluster centers (centroids)

Iteration until no changeIteration until no change For each object Oi

Calculate the distances between Oi and the k centroids

(Re)assign Oi to the cluster whose centroid is the closest to Oi

Update the cluster centroids based on current assignment

27

CSE300

kk-Means Clustering Method -Means Clustering Method

0

1

2

3

4

5

6

7

8

9

10

0 1 2 3 4 5 6 7 8 9 10

0

1

2

3

4

5

6

7

8

9

10

0 1 2 3 4 5 6 7 8 9 10

0

1

2

3

4

5

6

7

8

9

10

0 1 2 3 4 5 6 7 8 9 10

0

1

2

3

4

5

6

7

8

9

10

0 1 2 3 4 5 6 7 8 9 10

cluster

meancurrent clusters

new clusters

objectsrelocat

ed

28

CSE300

Dataset Dataset Data set from UCI repositoryData set from UCI repository http://kdd.ics.uci.edu/http://kdd.ics.uci.edu/ 768 female Pima Indians evaluated for diabetes768 female Pima Indians evaluated for diabetes After data cleaning 392 data entriesAfter data cleaning 392 data entries

29

CSE300

Hierarchical ClusteringHierarchical Clustering Groups observations based on dissimilarityGroups observations based on dissimilarity Compacts database into “labels” that represent the Compacts database into “labels” that represent the

observationsobservations Measure of similarity/DissimilarityMeasure of similarity/Dissimilarity

Euclidean Distance Manhattan Distance

Types of ClusteringTypes of Clustering Single Link Average Link Complete Link

30

CSE300

Hierarchical Clustering: ComparisonHierarchical Clustering: Comparison

Average-link Centroid distance

1

2

3

4

5

61

2

5

3

4

Single-link Complete-link

1

2

3

4

5

61

2

5

34

1

2

3

4

5

61

2 5

3

41

2

3

4

5

6

12

3

4

5

31

CSE300

Compare DendrogramsCompare Dendrograms

1 2 5 3 6 4 1 2 5 3 6 4

1 2 5 3 6 4 2 5 3 6 4 1

Average-link Centroid distance

Single-link Complete-link

32

CSE300

Which Distance Measure is Better?Which Distance Measure is Better? Each method has both advantages and disadvantages; Each method has both advantages and disadvantages;

application-dependentapplication-dependent Single-linkSingle-link

Can find irregular-shaped clusters Sensitive to outliers

Complete-link, Average-link, and Centroid distanceComplete-link, Average-link, and Centroid distance Robust to outliers Tend to break large clusters Prefer spherical clusters

33

CSE300

Dendrogram from datasetDendrogram from dataset

Minimum spanning tree through the observationsMinimum spanning tree through the observations Single observation that is last to join the cluster is patient whose Single observation that is last to join the cluster is patient whose

blood pressure is at bottom quartile, skin thickness is at bottom blood pressure is at bottom quartile, skin thickness is at bottom quartile and BMI is in bottom halfquartile and BMI is in bottom half

Insulin was however largest and she is 59-year old diabeticInsulin was however largest and she is 59-year old diabetic

34

CSE300

Dendrogram from datasetDendrogram from dataset

Maximum dissimilarity between observations in one Maximum dissimilarity between observations in one cluster when compared to anothercluster when compared to another

35

CSE300

Dendrogram from datasetDendrogram from dataset

Average dissimilarity between observations in one Average dissimilarity between observations in one cluster when compared to anothercluster when compared to another

36

CSE300

Supervised versus Unsupervised LearningSupervised versus Unsupervised Learning Supervised learning (classification)Supervised learning (classification)

Supervision: Training data (observations, measurements, etc.) are accompanied by labels indicating the class of the observations

New data is classified based on training set Unsupervised learning (clustering)Unsupervised learning (clustering)

Class labels of training data are unknown Given a set of measurements, observations, etc.,

need to establish existence of classes or clusters in data

37

CSE300

Derive models that can use patient specific Derive models that can use patient specific information, aid clinical decision makinginformation, aid clinical decision making

Apriori decision on predictors and variables to predictApriori decision on predictors and variables to predict No method to find predictors that are not present in the No method to find predictors that are not present in the

datadata Numeric ResponseNumeric Response

Least Squares Regression Categorical ResponseCategorical Response

Classification trees Neural Networks Support Vector Machine

Decision modelsDecision models Prognosis, Diagnosis and treatment planning Embed in clinical information systems

Classification and PredictionClassification and Prediction

38

CSE300

Least Squares RegressionLeast Squares Regression Find a linear function of predictor variables that Find a linear function of predictor variables that

minimize the sum of square difference with response minimize the sum of square difference with response Supervised learning techniqueSupervised learning technique

Predict insulin in our dataset :glucose and BMIPredict insulin in our dataset :glucose and BMI

39

CSE300

Decision TreesDecision Trees Decision treeDecision tree

Each internal node tests an attribute Each branch corresponds to attribute value Each leaf node assigns a classification

ID3 algorithmID3 algorithm Based on training objects with known class labels to

classify testing objects Rank attributes with information gain measure Minimal height

least number of tests to classify an object Used in commercial tools eg: Clementine ASSISTANT

Deal with medical datasets Incomplete data Discretize continuous variables Prune unreliable parts of tree Classify data

40

CSE300

Decision TreesDecision Trees

41

CSE300

Algorithm for Decision Tree InductionAlgorithm for Decision Tree Induction

Basic algorithm (a greedy algorithm)Basic algorithm (a greedy algorithm) Attributes are categorical (if continuous-valued,

they are discretized in advance) Tree is constructed in a top-down recursive

divide-and-conquer manner At start, all training examples are at the root Test attributes are selected on basis of a heuristic

or statistical measure (e.g., information gain) Examples are partitioned recursively based on

selected attributes

42

CSE300

Training DatasetTraining Dataset

AgeAge BMIBMI HereditaryHereditary VisionVision Risk of Risk of Condition XCondition X

P1P1 <=30<=30 highhigh nono fairfair nono

P2P2 <=30<=30 highhigh nono excellentexcellent nono

P3P3 >40>40 highhigh nono fairfair yesyes

P4P4 31…4031…40 mediummedium nono fairfair yesyes

P5P5 31…4031…40 lowlow yesyes fairfair yesyes

P6P6 31…4031…40 lowlow yesyes excellentexcellent nono

P7P7 >40>40 lowlow yesyes excellentexcellent yesyes

P8P8 <=30<=30 mediummedium nono fairfair nono

P9P9 <=30<=30 lowlow yesyes fairfair yesyes

P10P10 31…4031…40 mediummedium yesyes fairfair yesyes

P11P11 <=30<=30 mediummedium yesyes excellentexcellent yesyes

P12P12 >40>40 mediummedium nono excellentexcellent yesyes

P13P13 >40>40 highhigh yesyes fairfair yesyes

P14P14 31…4031…40 mediummedium nono excellentexcellent nono

43

CSE300

Construction of A Decision Tree for “Condition X”Construction of A Decision Tree for “Condition X”

Age?

>4030…40

<=30

[P1,…P14]Yes: 9,

No:5

[P1,P2,P8,P9,P11]

Yes: 2, No:3

[P3,P7,P12,P13]Yes: 4, No:0

[P4,P5,P6,P10,P14]

Yes: 3, No:2History

no yes

YES

[P1,P2,P8]

Yes: 0, No:3

[P9,P11]Yes: 2,

No:0

Vision

fairexcellent

NO YES NO YES

[P6,P14]Yes: 0,

No:2

[P4,P5,P10]

Yes: 3, No:0

44

CSE300

Entropy and Information GainEntropy and Information Gain SS contains contains ssii tuples of class tuples of class CCii for for ii = {1, ..., = {1, ..., mm} } Information measures info required to classify any Information measures info required to classify any

arbitrary tuplearbitrary tuple

Entropy of attribute A with values {aEntropy of attribute A with values {a11,a,a22,…,a,…,avv}}

Information gained by branching on attribute AInformation gained by branching on attribute A

ss

ss

,...,s,ssi

m

i

im21 2

1

log)I(

),...,(...

E(A) 11

1

mjjmjj

ssIs

ssv

j

)E(),...,,I()Gain( 21 AsssA m

45

CSE300

Entropy and Information GainEntropy and Information Gain Select attribute with the Select attribute with the highesthighest information gain (or information gain (or

greatest entropy reduction)greatest entropy reduction) Such attribute minimizes information needed to

classify samples

46

CSE300

Rule InductionRule Induction IF conditions THEN ConclusionIF conditions THEN Conclusion Eg: CN2Eg: CN2

Concept description: Characterization: provides a concise and succinct summarization

of given collection of data Comparison: provides descriptions comparing two or more

collections of data

Training set, testing setTraining set, testing set ImpreciseImprecise Predictive AccuracyPredictive Accuracy

P/P+N

47

CSE300

Example used in a ClinicExample used in a Clinic Hip arthoplasty trauma surgeon predict patient’s long-Hip arthoplasty trauma surgeon predict patient’s long-

term clinical status after surgeryterm clinical status after surgery Outcome evaluated during follow-ups for 2 yearsOutcome evaluated during follow-ups for 2 years 2 modeling techniques2 modeling techniques

Naïve Bayesian classifier Decision trees

Bayesian classifierBayesian classifier P(outcome=good) = 0.55 (11/20 good) Probability gets updated as more attributes are

considered P(timing=good|outcome=good) = 9/11 (0.846) P(outcome = bad) = 9/20 P(timing=good|

outcome=bad) = 5/9

48

CSE300

NomogramNomogram

49

CSE300

Bayesian ClassificationBayesian Classification Bayesian classifier vs. decision treeBayesian classifier vs. decision tree

Decision tree: predict the class label Bayesian classifier: statistical classifier; predict

class membership probabilities Based on Based on Bayes theoremBayes theorem; estimate ; estimate posteriorposterior

probabilityprobability Naïve Bayesian classifier: Naïve Bayesian classifier:

Simple classifier that assumes attribute independence

High speed when applied to large databases Comparable in performance to decision trees

50

CSE300

Bayes TheoremBayes Theorem Let Let XX be a data sample whose class label is unknown be a data sample whose class label is unknown Let Let HHii be the hypothesis that be the hypothesis that XX belongs to a particular belongs to a particular

class class CCii

P(P(HHii) is ) is class priorclass prior probability that probability that XX belongs to a belongs to a particular class particular class CCii

Can be estimated by ni/n from training data samples

n is the total number of training data samples ni is the number of training data samples of class Ci

)()()|(

)|(XP

iHPiHXPXiHP

Formula of Bayes Theorem

51

CSE300

More classification TechniquesMore classification Techniques Neural NetworksNeural Networks

Similar to pattern recognition properties of biological systems

Most frequently used Multi-layer perceptrons

– Input with bias, connected by weights to hidden, output

Backpropagation neural networks Support Vector MachinesSupport Vector Machines

Separate database to mutually exclusive regions Transform to another problem space Kernel functions (dot product) Output of new points predicted by position

Comparison with classification treesComparison with classification trees Not possible to know which features or combination of

features most influence a prediction

52

CSE300

Multilayer PerceptronsMultilayer Perceptrons Non-linear transfer functions to weighted sums of Non-linear transfer functions to weighted sums of

inputsinputs Werbos algorithmWerbos algorithm

Random weights Training set, Testing set

53

CSE300

Support Vector MachinesSupport Vector Machines 3 steps3 steps

Support Vector creation Maximal distance between points found Perpendicular decision boundary

Allows some points to be misclassifiedAllows some points to be misclassified Pima Indian data with X1(glucose) X2(BMI)Pima Indian data with X1(glucose) X2(BMI)

54

CSE300

What is Association Rule Mining?What is Association Rule Mining? Finding frequent patterns, associations, correlations, or causal Finding frequent patterns, associations, correlations, or causal

structures among sets of items or objects in transaction structures among sets of items or objects in transaction databases, relational databases, and other information databases, relational databases, and other information repositoriesrepositories

Example of Association Rules{High LDL, Low HDL} {Heart Failure}

PatientIDPatientID ConditionsConditions11 High LDL Low HDL, High LDL Low HDL,

High BMI, High BMI, Heart FailureHeart Failure

22 High LDL Low HDLHigh LDL Low HDL, , Heart Failure, Heart Failure, DiabetesDiabetes

33 DiabetesDiabetes

44 High LDL Low HDLHigh LDL Low HDL, , Heart FailureHeart Failure

55 High BMIHigh BMI , , High LDL High LDL Low HDLLow HDL, , Heart FailureHeart Failure

People who have high LDL (“bad” cholesterol), low HDL (“good cholesterol”) are at

higher risk of heart failure.

55

CSE300

Association Rule MiningAssociation Rule Mining Market Basket AnalysisMarket Basket Analysis

Same groups of items bought placed together Healthcare

Understanding among association among patients with demands for similar treatments and services

Goal : find items for which joint probability of occurrence is high

Basket of binary valued variables Results form association rules, augmented with

support and confidence

56

CSE300

Association Rule MiningAssociation Rule Mining

Dintrans

YXcontainingtransYXP

#

)(#)(

Association RuleAssociation Rule An implication

expression of the form X Y, where X and Y are itemsets and XY=

Rule Evaluation Rule Evaluation MetricsMetrics Support (s): Fraction of

transactions that contain both X and Y

Confidence (c): Measures how often items in Y appear in transactions thatcontain X

Xcontainingtrans

YXcontainingtransYXP

#

)(#)|(

Trans containing Y

Trans containing

both X and Y

Trans containing X

D

57

CSE300

The Apriori AlgorithmThe Apriori Algorithm Starts with most frequent 1-itemsetStarts with most frequent 1-itemset Include only those “items” that pass thresholdInclude only those “items” that pass threshold Use 1-itemset to generate 2-itemsetsUse 1-itemset to generate 2-itemsets Stop when threshold not satisfied by any itemsetStop when threshold not satisfied by any itemset

LL11 = {frequent items}; = {frequent items};for (k = 1; for (k = 1; LLkk != !=; k++) do; k++) do Candidate Generation: Ck+1 = candidates

generated from Lk; Candidate Counting: for each transaction t in

database do increment the count of all candidates in Ck+1 that are contained in t

Lk+1 = candidates in Ck+1 with min_supreturn return k k LLkk;;

58

CSE300

Apriori-based MiningApriori-based Mining

b, eb, e4040

a, b, c, ea, b, c, e3030

b, c, eb, c, e2020

a, c, da, c, d1010

ItemsItemsTIDTID

Min_sup=0.511dd

33ee

33cc

33bb

22aa

SupSupItemsetItemsetData base D 1-candidates

Scan D

33ee

33cc

33bb

22aa

SupSupItemsetItemsetFreq 1-itemsets

bcbc

aeae

acac

cece

bebe

abab

ItemsetItemset2-candidates

cece

bebe

bcbc

aeae

acac

abab

ItemsetItemset

22

11

22

22

33

11

SupSup

Counting

Scan D

cece

bebe

bcbc

acac

ItemsetItemset

22

22

22

33

SupSup

Freq 2-itemsets

bcebce

ItemsetItemset

3-candidates

bcebce

ItemsetItemset

22

SupSup

Freq 3-itemsets

Scan D

59

CSE300

Principle Component AnalysisPrinciple Component Analysis Principle ComponentsPrinciple Components

In cases of large number of variables, highly possible that some subsets of the variables are very correlated with each other. Reduce variables but retain variability in dataset

Linear combinations of variables in the database Variance of each PC maximized

– Display as much spread of the original data PC orthogonal with each other

– Minimize the overlap in the variables Each component normalized sum of square is unity

– Easier for mathematical analysis Number of PC < Number of variables

Associations found Small number of PC explain large amount of variance

Example 768 female Pima Indians evaluated for diabetes Number of times pregnant, two-hour oral glucose tolerance test

(OGTT) plasma glucose, Diastolic blood pressure, Triceps skin fold thickness, Two-hour serum insulin, BMI, Diabetes pedigree function, Age, Diabetes onset within last 5 years

60

CSE300

PCA ExamplePCA Example

61

CSE300

National Cancer InstituteNational Cancer Institute CancerNet CancerNet http://www.nci.nih.govhttp://www.nci.nih.gov CancerNet for Patients and the PublicCancerNet for Patients and the Public CancerNet for Health ProfessionalsCancerNet for Health Professionals CancerNet for Basic ReasearchersCancerNet for Basic Reasearchers CancerLitCancerLit

62

CSE300

ConclusionConclusion About ¾ billion of people’s medical records are About ¾ billion of people’s medical records are

electronically availableelectronically available Data mining in medicine distinct from other fields due Data mining in medicine distinct from other fields due

to nature of data: heterogeneous, with ethical, legal to nature of data: heterogeneous, with ethical, legal and social constraintsand social constraints

Most commonly used technique is classification and Most commonly used technique is classification and prediction with different techniques applied for prediction with different techniques applied for different casesdifferent cases

Associative rules describe the data in the databaseAssociative rules describe the data in the database Medical data mining can be the most rewarding Medical data mining can be the most rewarding

despite the difficultydespite the difficulty

63

CSE300

Thank you !!!Thank you !!!