Decomposition Methodology For Knowledge Discovery And Data Mining: Theory And Applications

Note: Supplemental materials are not guaranteed with Rental or Used book purchases.

ISBN: 9789812560797 | 9812560793
Cover: Hardcover
Copyright: 5/30/2005

Rent

(Recommended)

$87.33

Term

Due

Price

Semester

Dec 12

$91.93

Quarter

Aug 13

$87.33

Short Term

Jul 28

$82.74

*This item is part of an exclusive publisher rental program and requires an additional convenience fee. This fee will be reflected in the shopping bag.
Buy New

In Stock Usually Ships in 24 Hours

$117.69
eBook

Available Instantly

Online: 1825 Days

Downloadable: Lifetime Access

*To support the delivery of the digital material to you, a digital delivery fee of $3.99 will be charged on each digital item.

$57.60*

Data Mining is the science and technology of exploring data in order to discover previously unknown patterns. It is a part of the overall process of Knowledge Discovery from Databases (KDD). The accessibility and abundance of information today makes data mining a matter of considerable importance and necessity. This book provides an introduction to the field with an emphasis on advanced decomposition methods in general data mining tasks and for classification tasks in particular. The book presents a complete methodology for decomposing classification problems into smaller and more manageable sub-problems that are solvable by using existing tools, and then joining them together to solve the initial problem. The benefits of decomposition methodology in data mining include: increased performance (classification accuracy), conceptual simplification of the problem, enhanced feasibility with huge databases, clearer and more comprehensible results, reduced runtime by solving smaller problems and by using parallel/distributed computation, and the opportunity of using different techniques for individual sub-problems.

Preface

vii

Knowledge Discovery and Data Mining: Concepts and Fundamental Aspects

(34)

Overview

(1)

Data Mining and Knowledge Discovery

(2)

Taxonomy of Data Mining Methods

(1)

Supervised Methods

(5)

Overview

(1)

Training Set

(1)

Definition of the Classification Problem

(2)

Induction Algorithms

(1)

Rule Induction

(1)

Decision Trees

(2)

Bayesian Methods

(4)

Overview

(1)

Naive Bayes

(1)

The Basic Naive Bayes Classifier

(2)

Naive Bayes for Numeric Attributes

(1)

Correction to the Probability Estimation

(1)

Laplace Correction

(1)

No Match

(1)

Other Bayesian Methods

(1)

Other Induction Methods

(4)

Neural Networks

(2)

Genetic Algorithms

(1)

Instancebased Learning

(1)

Support Vector Machines

(1)

Performance Evaluation

(8)

Generalization Error

(1)

Theoretical Estimation of Generalization Error

(1)

VC-Framework

(2)

PAC-Framework

(1)

Empirical Estimation of Generalization Error

(2)

Bias and Variance Decomposition

(1)

Computational Complexity

(1)

Comprehensibility

(1)

``No Free Lunch'' Theorem

(2)

Scalability to Large Datasets

(2)

The ``Curse of Dimensionality''

(3)

Decision Trees

(26)

Decision Trees

(2)

Algorithmic Framework for Decision Trees

(1)

Univariate Splitting Criteria

(6)

Overview

(1)

Impurity Based Criteria

(2)

Information Gain

(1)

Gini Index

(1)

Likelihood Ratio Chi-Squared Statistics

(1)

DKM Criterion

(1)

Normalized Impurity Based Criteria

(1)

Gain Ratio

(1)

Distance Measure

(1)

Binary Criteria

(1)

Twoing Criterion

(1)

Orthogonal Criterion

(1)

Kolmogorov-Smirnov Criterion

(1)

AUC Splitting Criteria

(1)

Other Univariate Splitting Criteria

(1)

Comparison of Univariate Splitting Criteria

(1)

Multivariate Splitting Criteria

(1)

Stopping Criteria

(1)

Pruning Methods

(5)

Overview

(1)

Cost-Complexity Pruning

(1)

Reduced Error Pruning

(1)

Minimum Error Pruning (MEP)

(1)

Pessimistic Pruning

(1)

Error-Based Pruning (EBP)

(1)

Optimal Pruning

(1)

Minimum Description Length Pruning

(1)

Other Pruning Methods

(1)

Comparison of Pruning Methods

(1)

Other Issues

(2)

Weighting Instances

(1)

Misclassification costs

(1)

Handling Missing Values

(2)

Decision Trees Inducers

(9)

ID3

(1)

C4.5

(1)

CART

(1)

CHAID

(1)

QUEST

(1)

Reference to Other Algorithms

(1)

Advantages and Disadvantages of Decision Trees

(2)

Oblivious Decision Trees

(1)

Decision Trees Inducers for Large Datasets

(2)

Incremental Induction

(2)

Clustering Methods

(34)

Introduction

(1)

Distance Measures

(3)

Minkowski: Distance Measures for Numeric Attributes

(1)

Distance Measures for Binary Attributes

(1)

Distance Measures for Nominal Attributes

(1)

Distance Metrics for Ordinal Attributes

(1)

Distance Metrics for Mixed-Type Attributes

(1)

Similarity Functions

(1)

Cosine Measure

(1)

Pearson Correlation Measure

(1)

Extended Jaccard Measure

(1)

Dice Coefficient Measure

(1)

Evaluation Criteria Measures

(6)

Internal Quality Criteria

(1)

Sum of Squared Error (SSE)

(1)

Other Minimum Variance Criteria

(1)

Scatter Criteria

(1)

Condorcet's Criterion

(1)

The C-Criterion

(1)

Category Utility Metric

(1)

Edge Cut Metrics

(1)

External Quality Criteria

(1)

Mutual Information Based Measure

(1)

Precision-Recall Measure

(1)

Rand Index

(1)

Clustering Methods

(14)

Hierarchical Methods

(3)

Partitioning Methods

(1)

Error Minimization Algorithms

(2)

Graph-Theoretic Clustering

(1)

Density Based Methods

(2)

Model Based Clustering

(1)

Decision Trees

(1)

Neural Networks

(1)

Grid-Based Methods

(1)

Fuzzy Clustering

(1)

Evolutionary Approaches for Clustering

(3)

Simulated Annealing for Clustering

(1)

Which Technique to Use?

(2)

Clustering Large Data Sets

(4)

Decomposition Approach

(1)

Incremental Clustering

(2)

Parallel Implementation

(1)

Determining the Number of Clusters

(5)

Methods Based on Intra Cluster Scatter

(1)

Methods Based on Both the Inter and Intra Cluster Scatter

(2)

Criteria Based on Probabilistic

(1)

Ensemble Methods

(28)

Multiple Classifiers

(1)

Ensemble Methodology

(1)

Sequential Methodology

(8)

Model Guided Instance Selection

(1)

Uncertainty Sampling

(1)

Boosting

100

(3)

Windowing

103

(2)

Incremental Batch Learning

105

(1)

Concurrent Methodology

105

(3)

Bagging

106

(2)

Cross-validated Committees

108

(1)

Combining Classifiers

108

(8)

Simple Combining Methods

108

(1)

Uniform Voting

108

(1)

Distribution Summation

109

(1)

Bayesian Combination

109

(1)

Dempster-Shafer

109

(1)

Naive Bayes

110

(1)

Entropy Weighting

110

(1)

Density Based Weighting

110

(1)

DEA Weighting Method

111

(1)

Logarithmic Opinion Pool

111

(1)

Order Statistics

111

(1)

Meta Combining Methods

111

(1)

Stacking

111

(1)

Arbiter Trees

112

(3)

Combiner Trees

115

(1)

Grading

115

(1)

Ensemble Size

116

(2)

Selecting the Ensemble Size

116

(1)

Pruning Ensembles

117

(1)

Ensemble Diversity

118

(5)

Manipulating the Inducer

118

(1)

Manipulating the Training Set

119

(1)

Manipulating the Tuples

119

(1)

Manipulating the Input Feature Set

119

(2)

Manipulating the Target Attribute

121

(1)

Measuring the Diversity

121

(2)

Elementary Decomposition Framework

123

(28)

Decomposition Framework

123

(2)

Decomposition Advantages

125

(3)

Increasing Classification Performance (Classification Accuracy)

125

(2)

Scalability to Large Databases

127

(1)

Increasing Comprehensibility

127

(1)

Modularity

127

(1)

Suitability for Parallel Computation

128

(1)

Flexibility in Techniques Selection

128

(1)

The Elementary Decomposition Methodology

128

(23)

Illustrative Example of Elementary Decomposition

131

(1)

Overview

131

(1)

Feature Set Decomposition

132

(1)

Sample Decomposition

133

(1)

Space Decomposition

133

(1)

Concept Aggregation

134

(1)

Function Decomposition

135

(1)

The Decomposer's Characteristics

136

(1)

Overview

136

(1)

The Structure Acquiring Method

137

(1)

The Mutually Exclusive Property

138

(1)

The Inducer Usage

139

(1)

Exhaustiveness

140

(1)

Combiner Usage

141

(1)

Sequentially or Concurrently

141

(1)

Distributed and Parallel Data Mining

142

(1)

The Uniqueness of the Proposed Decomposition Framework

143

(2)

Problems Formulation

145

(1)

Feature Set Decomposition

146

(1)

Space Decomposition

147

(1)

Sample Decomposition

148

(1)

Concept Aggregation

148

(1)

Function Decomposition

149

(2)

Feature Set Decomposition

151

(50)

Overview

151

(4)

Problem Formulation

155

(2)

Definitions and Properties

157

(3)

Conditions for Complete Equivalence

160

(7)

Algorithmic Framework for Feature Set Decomposition

167

(17)

Overview

167

(1)

Searching Methods

168

(2)

Induction Methods

170

(4)

Accuracy Evaluation Methods

174

(1)

The Wrapper Method

174

(1)

Conditional Entropy

174

(1)

VC-Dimension

175

(6)

Classification of an Unlabeled Instance

181

(1)

Computational Complexity

181

(2)

Specific Algorithms Implementations

183

(1)

DOG

183

(1)

BCW

184

(1)

Experimental Study

184

(17)

Overview

184

(1)

Algorithms Used

184

(1)

Data Sets Used

185

(2)

Metrics Measured

187

(1)

Comparison with Single Model Algorithms

188

(1)

Comparing to Ensemble Algorithms

189

(2)

Discussion

191

(1)

The Relation of Feature Set Decomposition Performance and Node-Sample Ratio

191

(1)

The Link Between Error Reduction and the Problem Complexity

192

(1)

The Suitability of the Naive Bayes Combination

193

(1)

The Effect of Subset Size

193

(8)

Space Decomposition

201

(14)

Overview

201

(1)

Motivation

201

(3)

Problem Formulation

204

(6)

Manners for Dividing the Instance Space

205

(1)

The K-Means Algorithm as a Decomposition Tool

206

(1)

Determining the Number of Subsets

207

(1)

The Basic K-Classifier Algorithm

208

(2)

The Heterogeneity Detecting K-Classifier (HDK-Classifier)

210

(1)

Overview

210

(1)

Running-Time Complexity

211

(1)

Experimental Study

211

(1)

Conclusions

212

(3)

Sample Decomposition

215

(16)

Overview

215

(1)

Tuple Sampling

215

(1)

Problem Formulation

216

(1)

The CBCD Algorithm

217

(3)

Stages of the Algorithm

218

(1)

Stage 1 --- Apply the K-Means Algorithm on the Entire DataSet S

218

(1)

Stage 2 --- Produce the ψ Subsets Based on the K Clusters Created Sn Stage 1

218

(1)

Stage 3 --- Produce ψ Classifiers by Applying the C4.5 Learning Algorithm on the ψ Cluster-Based Subsets

219

(1)

Running-Time Complexity Analysis

219

(1)

Classifying New Instances

220

(1)

An Illustrative Example

220

(4)

Step 1 - Creating a Training Set

221

(1)

Step 2 - Performing Clustering

222

(1)

Step 3 - Creating the Subsets

222

(1)

Step 4 - Performing Induction

222

(1)

Step 5 - Testing the Classifiers

223

(1)

Experimental Study

224

(5)

Data Sets Used

225

(1)

Experimental Stages

225

(1)

Stage 1---Partitioning the Dataset

225

(1)

Stage 2---Applying the CBCD Algorithm on the Datasets

226

(1)

Stage 3---Employing the Paired T Test

226

(3)

Conclusions

229

(2)

Function Decomposition

231

(28)

Data Preprocessing

231

(1)

Feature Transformation

232

(4)

Feature Aggregation

234

(2)

Supervised and Unsupervised Discretization

236

(1)

Function Decomposition

236

(5)

Boolean Function Decomposition in Switching Circuits

236

(1)

Feature Discovery by Constructive Induction

237

(1)

Function Decomposition in Data Mining

238

(3)

Problem Formulation

241

(1)

The IBAC Algorithm

241

(8)

The Basic Stage

243

(1)

Initialization

243

(1)

Classifier Comparison

244

(1)

Assigning Values to the New Attribute

244

(1)

Creating the New Dataset

245

(1)

Selection Criterion and Stopping Criterion

245

(1)

Time Complexity Analysis

246

(1)

Classifying New Instances

247

(1)

Example

247

(2)

Experimental Study

249

(3)

Databases Used

249

(1)

Dataset Discretization

249

(1)

Evaluating the Algorithm Performance

250

(1)

Experimental Results

250

(2)

Meta-Classifier

252

(5)

Building the Meta-Database

253

(1)

Inducing the Meta-Classifier

254

(1)

Examining the Meta-Learning Approach

254

(1)

Meta-Classifier Results

255

(2)

Conclusions

257

(2)

Concept Aggregation

259

(26)

Overview

259

(1)

Motivation

260

(1)

The Multi Criteria Problem

261

(1)

Problem Formulation

262

(1)

Basic Solution Approaches

263

(1)

Concept Aggregation---Similarity Based Approach (SBA)

264

(4)

SBA---Procedure

264

(1)

Meta-Data Vectors

265

(1)

SBA---Detailed Illustration

266

(2)

Concept Aggregation---Error Based Approach (eBA)

268

(8)

The Confusion Matrix

269

(1)

EBA Method---Optimal Solution

270

(3)

EBA Method---Heuristic Greedy Solution

273

(1)

Detailed Procedure

274

(1)

Illustration

275

(1)

Experimental Study

276

(3)

Solving the Optimization Problem of the EBA Optimal Method

276

(1)

Experiment Results

276

(3)

Concept De-Aggregation

279

(2)

Concept De-Aggregation ---Outline

280

(1)

Models Generation Procedure

280

(1)

Termination Condition

281

(1)

Finding the Optimal Partition

281

(1)

Conclusions

281

(4)

A Meta-Classification for Decomposition Methodology

285

(16)

Meta Classifier Schema

285

(2)

Dataset Characterizer

287

(3)

New Entropy Measure for Characterizing Mixed Datasets

290

(6)

Dataset Manipulator

296

(1)

Decomposer Evaluator

297

(1)

The Meta-Decomposer

298

(1)

Evaluating the Meta-Decomposer

298

(3)

Bibliography

301

(20)

Index

321

Please wait while the item is added to your bag...

FREE SHIPPING

Rent More, Save More! Use code: BBRENTAL

5% off 1 book, 7% off 2 books, 10% off 3+ books

Decomposition Methodology For Knowledge Discovery And Data Mining: Theory And Applications

Summary

Table of Contents

Supplemental Materials

FREE SHIPPING

Rent More, Save More! Use code: BBRENTAL

5% off 1 book, 7% off 2 books, 10% off 3+ books

Decomposition Methodology For Knowledge Discovery And Data Mining: Theory And Applications

Summary

Table of Contents

Supplemental Materials

Digital License