TABLE OF CONTENTS

CHAPTER NO.

TITLE

PAGE NO.

ABSTRACT
LIST OF TABLES
LIST OF FIGURES
LIST OF ABBREVIATIONS

iv
xii
xiv
xvii

INTRODUCTION
1.1 GENERAL
1.2 DATA MINING
1.3 ASSOCIATION RULE MINING
1.4 POPULATION BASED STOCHASTIC SEARCH

1
1
1
4
5

METHODS
1.5 PROBLEM IDENTIFICATION
1.6 OBJECTIVE OF THE RESEARCH
1.7 ORGANIZATION OF THE THESIS

7
8
9

LITERTURE REVIEW
2.1 GENERAL
2.2 TRADITIONAL METHODS FOR MINING

11
11
11

ASSOCIATION RULES
2.2.1 AIS Algorithm
2.2.2 Apriori and Apriori-Based Algorithms
2.2.3 FP-Tree and FP-Growth/FP-Tree Based

11
12
13

Algorithms
2.2.4 Other Association Rule Mining Methods
2.3 STOCHASTIC SEARCH METHODS FOR MINING

14
15

ASSOCIATION RULES
2.3.1 Genetic Algorithms for ARM
2.3.2 Particle Swarm Optimization for ARM
2.4 HYBRID APPROACHES OF GA AND PSO
2.5 MEMETIC PARTICLE SWARM OPTIMIZATION
2.6 CONCLUDING REMARKS

16
19
22
24
26

CONCEPTS AND PROBLEM SETTINGS

3.1 GENERAL

27
27

3.2 STOCHASTIC SEARCH METHODS

3.3 OVERVIEW OF EVOLUTIONARY COMPUTING
3.3.1 Genetic Algorithm
3.3.2 Particle Swarm Optimization
3.4 DATASETS USED IN THIS STUDY
3.5 PERFORMANCE EVALUATION
3.6 SUMMARY

27
29
32
35
38
39
40

ASSOCIATION RULE MINING USING

41

GENETIC ALGORITHM
4.1 GENERAL
4.2 NEED AND MOTIVATION
4.3 MINING AR USING GA
4.3.1 Experimental Setup
4.3.2 Results and Discussion
4.4 EFFECTIVE PARAMETER SETTING OF GA FOR

41
41
42
42
45
48

ARM
4.4.1 Methodology
4.4.2 Experimental Results and Discussion
4.5 GA WITH ELITISM FOR ARM
4.5.1 Elitism in GA
4.5.2 Experimental Results and Discussion
4.6 ADAPTIVE GA FOR ARM
4.6.1 Need for Adaption
4.6.2 Experimental Results and Discussion
4.7 SUMMARY

49
50
54
54
57
61
62
64
68

PARTICLE SWARM OPTIMIZATION FOR MINING

70

ASSOCIATION RULES
5.1 GENERAL
5.2 PSO FOR ARM
5.2.1 Methodology
5.2.2 Experimental Results and Discussion
5.3 CHAOTIC PSO FOR ARM
5.3.1 Chaotism in PSO
5.3.2 Experimental Results and Discussion
5.4 DYNAMIC NEIGHBORHOOD SELECTION IN PSO

70
70
71
74
78
78
80
81

FOR ARM
5.4.1 Dynamic Neighborhood Selection in PSO
5.4.2 Experimental Results and Discussion

82
84

5.5 MOTIVATION FOR ADAPTIVE TECHNIQUES

5.6 DATA INEPENDENT ADAPTION IN PSO
5.6.1 Methodology
5.6.2 Experimental Results and Discussion
5.7 DATA DEPENDENT ADAPTION IN PSO
5.7.1 Methodology
5.7.2 Experimental Results and Discussion
5.8 SUMMARY
6

87
89
90
92
96
96
106
115

HYBRID AND MEMETIC METHOS FOR MINING 116

ASSOCIATION RULES
6.1 GENERAL
6.2 NEED FOR HYBRIDIZATION
6.3 HYBRID OF GA AND PSO FOR ARM
6.3.1 GA-PSO Hybrid for ARM
6.3.2 Experimental Results and Discussion
6.4 PSO WITH LOCAL SEARCH
6.5 PSO WITH SFLA FOR MINING ASSOCIATION

116
116
116
117
121
127
128

RULES
6.5.1 Methodology
6.5.2 Experimental Results and Discussion
6.6 SUMMARY

129
131
137

CONCLUSIONS
7.1 GENERAL
7.2 SALIENT CONCLUSIONS
7.3 SCOPE FOR FUTURE WORK

138
138
138
140

REFERENCES
LIST OF PUBLICATIONS
CURRICULUM VITAE
ANNEXURE I

141
160
167
170

LIST OF TABLES
TABLE
NO.
3.1

TITLE

Description of the Datasets Used

PAGE
NO.
38

4.1
4.2
4.3
4.4
4.5
4.6
4.7
4.8
4.9
4.10
4.11
4.12
4.13
5.1
5.2
5.3
5.4
5.5
5.6
5.7
5.8

GA Parameters for ARM

Predictive Accuracy for ARM with GA
Number of Rules Generated by GA
Qualitative Measures of AR mined by GA
Role of GA Parameters in Problem Solving
Default GA Parameters
Comparison based on variation in Crossover Probability for
ARM using GA
Default Parameters for ARM with Elitist GA
Predictive Accuracy for AR mined with Elitist GA
Qualitative Measures of ARM with Elitist GA
Default GA Parameters for ARM
Qualitative Measures of ARM by AGA
Comparison of PA achieved by the GA based Methods
Initial Parameter Setting of PSO for ARM
Predictive Accuracy comparison of PSO and GA for ARM
Qualitative Measures of ARM with PSO
Default Parameter setting of NPSO for ARM
Measure of Interestingness for Dynamic Neighborhood
Selection PSO
Comparison of Convergence rate of NPSO with PSO for
ARM
Parameters of PSO and their roles
Default Parameter Setting of Adaptive methodologies for

45
46
46
47
49
50
52
57
58
60
65
67
68
74
74
76
84
85
87
88
92

ARM
TABLE
NO.

TITLE

PAGE
NO.

5.9
5.10
5.11
5.12
5.13

Classification into States by Evolutionary Factor

Control Strategies for c1 and c2
Parameter Setting for APSO
Multiobjective Measures of APSO
Predictive Accuracy Comparison of APSO with AGA and

100
102
106
108
112

5.14
6.1
6.2
6.3

ACO
Predictive Accuracy Comparison of PSO based Methods
Parameter values of GPSO for ARM
Comparison of Predictive Accuracy
Comparison of the No. of Rules Generated in ARM

114
121
136
136

LIST OF FIGURES
FIGUR

TITLE

PAGE

E NO.
3.1
3.2
3.3
3.4
3.5
4.1
4.2
4.3

Block Diagram of Genetic Algorithm

Pseudo-code for simple GA
Crossover Operators
Pseudo-code for simple PSO
Displacement of a PSO particle
Roulette Wheel Selection Methodology
Population Size vs Accuracy for ARM with GA
Minimum Support and Minimum Confidence vs Accuracy

4.4
4.5

for ARM with GA

Elitism in Multiobjective Optimization
Block Diagram of Elitist GA for Association

55
56

4.6
4.7

Rule Mining
Flow chart of ARM with Elitist GA
Predictive Accuracy for Mining AR based on GA

57
59

with Elitism

NO.
33
34
35
36
37
44
51
53

4.8
4.9

Comparison of Elites for ARM with and without Elitism

Comparison of No. of Rules Generated with Elitist GA

60
61

4.10

and Simple GA
Comparison of the Accuracy between GA and AGA when

66

4.11

Parameters are Ideal for Traditional GA

Predictive Accuracy Comparison of AGA , GA and GA with

66

5.1
5.2
5.3

AGA Mutation Rate

Particles Movement towards the target
Flowchart of PSO for ARM
Comparison of Execution Time of PSO and GA for ARM

71
72
75

FIGUR

TITLE

PAGE

E NO.
5.4
5.5

Impact of Inertia Weight on Predictive Accuracy for ARM

Comparison of PSO with and without Inertia Weight

5.6

for ARM
Comparison of Predictive Accuracy of CPSO and PSO

80

5.7
5.8

for ARM
Convergence rate of CPSO for ARM
Predictive Accuracy Comparison for Dynamic

81
84

5.9
5.10

Neighborhood Selection in PSO

Execution Time Comparison of NPSO with PSO
Comparison of Predictive Accuracy of three Adaptive

86
93

5.11

Methodologies with PSO for ARM

Evolutionary State Information Robustly Revealed

99

5.12

by e at run time for Zoo dataset

Population Distribution in PSO based on Evolutionary

101

5.13

factor e
Adaptation of Acceleration Coefficients through EES

104

5.14
5.15

for Zoo Dataset

Inertia Weight Adaptation for Zoo dataset
Predictive Accuracy of Adaptive PSO over Hundred

105
107

5.16
5.17

Generations
Predictive Accuracy Comparison of APSO with PSO
Comparison of No. of Rules Generated by APSO and

109
109

PSO

NO.
77
77

5.18
5.19
5.20

FIGUR
E NO.
6.1
6.2

Comparison of Execution Time of APSO and PSO

Comparison of Convergence of APSO and PSO
Comparison of Convergence of APSO and SAPSO

TITLE

110
111
112

PAGE

Predictive Accuracy vs. No. of Generations for

NO.
119
122

6.3

ARM with GPSO

Comparison of Predictive Accuracy of GPSO with GA

125

6.4

and PSO
Comparison of Execution Time of GPSO with GA and

126

6.5
6.6
6.6
6.7

PSO
Flowchart for PSO with SFLA for ARM
Formation of Memeplexes
Predictive Accuracy Comparison for ARM
Fitness Value for PSO with SFLA for ARM

129
130
132
132

Hybrid GA/ PSO (GPSO) Model

LIST OF ABBREVIATIONS
ACO

Ant Colony Optimization

AGA

Adaptive Genetic Algorithm

AIS

Agrawal, Imielinski and Swami Algorithm

APSO

Adaptive Particle Swarm Optimization

APSO+SFLA -

Adaptive Particle Swarm Optimization with Shuffled Frog Leaping

Algorithm

AR

Association Rules

ARM

Association Rule Mining

BPSO

Binary Particle Swarm Optimization

CARMA

Continuous Association Rule Mining Algorithm

CPSO

Chaotic Particle Swarm Optimization

DIC

Dynamic Itemset Counting

EA

Evolutionary Algorithms

EC

Evolutionary Computing

EMA

Evolutionary Memetic Algorithm

EMOGA

Elitist Multi-Objective Genetic Algorithm

FP

Frequent Pattern

GA

Genetic Algorithm

GCPSO

Guaranteed Convergence Particle Swarm Optimization

GPSO

GA/PSO

ITL Mine

Item Trans Link Miner

KDD

Knowledge Discovery in Databases

LS

Local Search

MA

Memetic Algorithm

MPSO

Memetic Particle Swarm Optimization

NPSO

Neighborhood Selection in Particle Swarm Optimization

PA

Predictive Accuracy

PSO

Particle Swarm Optimization

Hybrid

PSO + SFLA -

Particle Swarm Optimization with Shuffled Frog Leaping

Algorithm

Q-QPSO

Quantum behaved Particle Swarm Optimization

RARM

Rapid Association Rule Mining

RWDE

Random Walk with Direction Exploitation

SAPSO1

Self Adaptive Particle Swarm Optimization 1

SAPSO2

Self Adaptive Particle Swarm Optimization 2

SACPSO

Self Adaptive Chaotic Particle Swarm Optimization

SFLA

Shuffled Frog Leaping Algorithm

SI

Swarm Intelligence

UCI

University of California, Irvine

VIPER

Vertical Itemset Partitioning for Efficient Rule

WPSO

Weighted Particle Swarm Optimization