Efnisyfirlit

• Preface
• Contents
• 1 Data Mining and Information Systems: Quo Vadis?
• Robert Stahlbock, Stefan Lessmann, and Sven F. Crone
• 1.1 Introduction
• 1.2 Special Issues in Data Mining
• 1.2.1 Confirmatory Data Analysis
• 1.2.2 Knowledge Discovery from Supervised Learning
• 1.2.3 Classification Analysis
• 1.2.4 Hybrid Data Mining Procedures
• 1.2.5 Web Mining
• 1.2.6 Privacy-Preserving Data Mining
• 1.3 Conclusion and Outlook
• References
• Part I Confirmatory Data Analysis
• 2 Response-Based Segmentation Using Finite Mixture Partial Least Squares
• Christian M. Ringle, Marko Sarstedt, and Erik A. Mooi
• 2.1 Introduction
• 2.1.1 On the Use of PLS Path Modeling
• 2.1.2 Problem Statement
• 2.1.3 Objectives and Organization
• 2.2 Partial Least Squares Path Modeling
• 2.3 Finite Mixture Partial Least Squares Segmentation
• 2.3.1 Foundations
• 2.3.2 Methodology
• 2.3.3 Systematic Application of FIMIX-PLS
• 2.4 Application of FIMIX-PLS
• 2.4.1 On Measuring Customer Satisfaction
• 2.4.2 Data and Measures
• 2.4.3 Data Analysis and Results
• 2.5 Summary and Conclusion
• References
• Part II Knowledge Discovery from Supervised Learning
• 3 Building Acceptable Classification Models
• David Martens and Bart Baesens
• 3.1 Introduction
• 3.2 Comprehensibility of Classification Models
• 3.2.1 Measuring Comprehensibility
• 3.2.2 Obtaining Comprehensible Classification Models
• 3.2.2.1 Building Rule-Based Models
• 3.2.2.2 Combining Output Types
• 3.2.2.3 Visualization
• 3.3 Justifiability of Classification Models
• 3.3.1 Taxonomy of Constraints
• 3.3.2 Monotonicity Constraint
• 3.3.3 Measuring Justifiability
• 3.3.4 Obtaining Justifiable Classification Models
• 3.4 Conclusion
• References
• 4 Mining Interesting Rules Without Support Requirement: A General Universal Existential Upward Closu
• Yannick Le Bras, Philippe Lenca, and Stéphane Lallich
• 4.1 Introduction
• 4.2 State of the Art
• 4.3 An Algorithmic Property of Confidence
• 4.3.1 On UEUC Framework
• 4.3.2 The UEUC Property
• 4.3.3 An Efficient Pruning Algorithm
• 4.3.4 Generalizing the UEUC Property
• 4.4 A Framework for the Study of Measures
• 4.4.1 Adapted Functions of Measure
• 4.4.1.1 Association Rules
• 4.4.1.2 Contingency Tables
• 4.4.2 Expression of a Set of Measures of Ddconf
• 4.5 Conditions for GUEUC
• 4.5.1 A Sufficient Condition
• 4.5.2 A Necessary Condition
• 4.5.3 Classification of the Measures
• 4.6 Conclusion
• References
• 5 Classification Techniques and Error Control in Logic Mining
• Giovanni Felici, Bruno Simeone, and Vincenzo Spinelli
• 5.1 Introduction
• 5.2 Brief Introduction to Box Clustering
• 5.3 BC-Based Classifier
• 5.4 Best Choice of a Box System
• 5.5 Bi-criterion Procedure for BC-Based Classifier
• 5.6 Examples
• 5.6.1 The Data Sets
• 5.6.2 Experimental Results with BC
• 5.6.3 Comparison with Decision Trees
• 5.7 Conclusions
• References
• Part III Classification Analysis
• 6 An Extended Study of the Discriminant Random Forest
• Tracy D. Lemmond, Barry Y. Chen, Andrew O. Hatch,and William G. Hanley
• 6.1 Introduction
• 6.2 Random Forests
• 6.3 Discriminant Random Forests
• 6.3.1 Linear Discriminant Analysis
• 6.3.2 The Discriminant Random Forest Methodology
• 6.4 DRF and RF: An Empirical Study
• 6.4.1 Hidden Signal Detection
• 6.4.1.1 Training on T1, Testing on J2
• 6.4.1.2 Prediction Performance for J2 with Cross-validation
• 6.4.2 Radiation Detection
• 6.4.3 Significance of Empirical Results
• 6.4.4 Small Samples and Early Stopping
• 6.4.5 Expected Cost
• 6.5 Conclusions
• References
• 7 Prediction with the SVM Using Test Point Margins
• Süreyya Özögür-Akyüz, Zakria Hussain, and John Shawe-Taylor
• 7.1 Introduction
• 7.2 Methods
• 7.3 Data Set Description
• 7.4 Results
• 7.5 Discussion and Future Work
• References
• 8 Effects of Oversampling Versus Cost-Sensitive Learning for Bayesian and SVM Classifiers
• Alexander Liu, Cheryl Martin, Brian La Cour, and Joydeep Ghosh
• 8.1 Introduction
• 8.2 Resampling
• 8.2.1 Random Oversampling
• 8.2.2 Generative Oversampling
• 8.3 Cost-Sensitive Learning
• 8.4 Related Work
• 8.5 A Theoretical Analysis of Oversampling Versus Cost-Sensitive Learning
• 8.5.1 Bayesian Classification
• 8.5.2 Resampling Versus Cost-Sensitive Learning in Bayesian Classifiers
• 8.5.3 Effect of Oversampling on Gaussian Naive Bayes
• 8.5.3.1 Random Oversampling
• 8.5.3.2 Generative Oversampling
• 8.5.3.3 Comparison to Cost-Sensitive Learning
• 8.5.4 Effects of Oversampling for Multinomial Naive Bayes
• 8.6 Empirical Comparison of Resampling and Cost-SensitiveLearning
• 8.6.1 Explaining Empirical Differences Between Resampling and Cost-Sensitive Learning
• 8.6.2 Naive Bayes Comparisons on Low-Dimensional Gaussian Data
• 8.6.2.1 Gaussian Naive Bayes on Artificial, Low-Dimensional Data
• 8.6.2.2 A Note on ROC and AUC
• 8.6.3 Multinomial Naive Bayes
• 8.6.4 SVMs
• 8.6.5 Discussion
• 8.7 Conclusion
• Appendix
• References
• 9 The Impact of Small Disjuncts on Classifier Learning
• Gary M. Weiss
• 9.1 Introduction
• 9.2 An Example: The Vote Data Set
• 9.3 Description of Experiments
• 9.4 The Problem with Small Disjuncts
• 9.5 The Effect of Pruning on Small Disjuncts
• 9.6 The Effect of Training Set Size on Small Disjuncts
• 9.7 The Effect of Noise on Small Disjuncts
• 9.8 The Effect of Class Imbalance on Small Disjuncts
• 9.9 Related Work
• 9.10 Conclusion
• References
• Part IV Hybrid Data Mining Procedures
• 10 Predicting Customer Loyalty Labels in a Large Retail Database: A Case Study in Chile
• Cristián J. Figueroa
• 10.1 Introduction
• 10.2 Related Work
• 10.3 Objectives of the Study
• 10.3.1 Supervised and Unsupervised Learning
• 10.3.2 Unsupervised Algorithms
• 10.3.2.1 Self-Organizing Map
• 10.3.2.2 Sammon Mapping
• 10.3.2.3 Curvilinear Component Analysis
• 10.3.3 Variables for Segmentation
• 10.3.4 Exploratory Data Analysis
• 10.3.5 Results of the Segmentation
• 10.4 Results of the Classifier
• 10.5 Business Validation
• 10.5.1 In-Store Minutes Charges for Prepaid Cell Phones
• 10.5.2 Distribution of Products in the Store
• 10.6 Conclusions and Discussion
• Appendix
• References
• 11 PCA-Based Time Series Similarity Search
• Leonidas Karamitopoulos, Georgios Evangelidis, and Dimitris Dervos
• 11.1 Introduction
• 11.2 Background
• 11.2.1 Review of PCA
• 11.2.2 Implications of PCA in Similarity Search
• 11.2.3 Related Work
• 11.3 Proposed Approach
• 11.4 Experimental Methodology
• 11.4.1 Data Sets
• 11.4.2 Evaluation Methods
• 11.4.3 Rival Measures
• 11.5 Results
• 11.5.1 1-NN Classification
• 11.5.2 k-NN Similarity Search
• 11.5.3 Speeding Up the Calculation of APEdist
• 11.6 Conclusion
• References
• 12 Evolutionary Optimization of Least-Squares Support Vector Machines
• Arjan Gijsberts, Giorgio Metta, and Léon Rothkrantz
• 12.1 Introduction
• 12.2 Kernel Machines
• 12.2.1 Least-Squares Support Vector Machines
• 12.2.2 Kernel Functions
• 12.2.2.1 Conditions for Kernels
• 12.3 Evolutionary Computation
• 12.3.1 Genetic Algorithms
• 12.3.2 Evolution Strategies
• 12.3.3 Genetic Programming
• 12.4 Related Work
• 12.4.1 Hyperparameter Optimization
• 12.4.2 Combined Kernel Functions
• 12.5 Evolutionary Optimization of Kernel Machines
• 12.5.1 Hyperparameter Optimization
• 12.5.2 Kernel Construction
• 12.5.3 Objective Function
• 12.6 Results
• 12.6.1 Data Sets
• 12.6.2 Results for Hyperparameter Optimization
• 12.6.3 Results for EvoKMGP
• 12.7 Conclusions and Future Work
• References
• 13 Genetically Evolved kNN Ensembles
• Ulf Johansson, Rikard König, and Lars Niklasson
• 13.1 Introduction
• 13.2 Background and Related Work
• 13.3 Method
• 13.3.1 Data sets
• 13.4 Results
• 13.5 Conclusions
• References
• Part V Web-Mining
• 14 Behaviorally Founded Recommendation Algorithm for Browsing Assistance Systems
• Peter Géczy, Noriaki Izumi, Shotaro Akaho, and Kôiti Hasida
• 14.1 Introduction
• 14.1.1 Related Works
• 14.1.2 Our Contribution and Approach
• 14.2 Concept Formalization
• 14.3 System Design
• 14.3.1 A Priori Knowledge of Human–System Interactions
• 14.3.2 Strategic Design Factors
• 14.3.3 Recommendation Algorithm Derivation
• 14.4 Practical Evaluation
• 14.4.1 Intranet Portal
• 14.4.2 System Evaluation
• 14.4.3 Practical Implications and Limitations
• 14.5 Conclusions and Future Work
• References
• 15 Using Web Text Mining to Predict Future Events: A Testof the Wisdom of Crowds Hypothesis
• Scott Ryan and Lutz Hamel
• 15.1 Introduction
• 15.2 Method
• 15.2.1 Hypotheses and Goals
• 15.2.2 General Methodology
• 15.2.3 The 2006 Congressional and Gubernatorial Elections
• 15.2.4 Sporting Events and Reality Television Programs
• 15.2.5 Movie Box Office Receipts and Music Sales
• 15.2.6 Replication
• 15.3 Results and Discussion
• 15.3.1 The 2006 Congressional and Gubernatorial Elections
• 15.3.2 Sporting Events and Reality Television Programs
• 15.3.3 Movie and Music Album Results
• 15.4 Conclusion
• References
• Part VI Privacy-Preserving Data Mining
• 16 Avoiding Attribute Disclosure with the (Extended) p-Sensitive k-Anonymity Model
• Traian Marius Truta and Alina Campan
• 16.1 Introduction
• 16.2 Privacy Models and Algorithms
• 16.2.1 The p-Sensitive k-Anonymity Model and Its Extension
• 16.2.2 Algorithms for the p-Sensitive k-Anonymity Model
• 16.3 Experimental Results
• 16.3.1 Experiments for p-Sensitive k-Anonymity
• 16.3.2 Experiments for Extended p-Sensitive k-Anonymity
• 16.4 New Enhanced Models Based on p-Sensitive k-Anonymity
• 16.4.1 Constrained p-Sensitive k-Anonymity
• 16.4.2 p-Sensitive k-Anonymity in Social Networks
• 16.5 Conclusions and Future Work
• References
• 17 Privacy-Preserving Random Kernel Classification of Checkerboard Partitioned Data
• Olvi L. Mangasarian and Edward W. Wild
• 17.1 Introduction
• 17.2 Privacy-Preserving Linear Classifier for Checkerboard Partitioned Data
• 17.3 Privacy-Preserving Nonlinear Classifier for Checkerboard Partitioned Data
• 17.4 Computational Results
• 17.5 Conclusion and Outlook
• References