Knowledge Discovery in Spatial Data

Acknowledgements

Preface

Contents

List of Figures

List of Tables

Introduction

1.1 On Spatial Data Mining and Knowledge Discovery

1.2 What Makes Spatial Data Mining Different

1.3 On Spatial Knowledge

1.4 On Spatial Data

1.5 Basic Tasks of Knowledge Discovery in Spatial Data

1.6 Issues of Knowledge Discovery in Spatial Data

1.7 Methodological Background for Knowledge Discovery in Spatial Data

1.8 Organization of the Book

Discovery of Intrinsic Clustering in Spatial Data

2.1 A Brief Background About Clustering

2.2 Discovery of Clustering in Space by Scale Space Filtering

2.2.1 On Scale Space Theory for Hierarchical Clustering

2.2.2 Hierarchical Clustering in Scale Space

2.2.3 Cluster Validity Check

2.2.4 Clustering Selection Rules

2.2.5 Some Numerical Examples

2.2.6 Discovering Land Covers in Remotely Sensed Images

2.2.7 Mining of Seismic Belts in Vector- Based Databases

2.2.8 Visualization of Temporal Seismic Activities via Scale Space Filtering

2.2.9 Summarizing Remarks on Clustering by Scale Space Filtering

2.3 Partitioning of Spatial Data by a Robust Fuzzy Relational Data Clustering Method

2.3.1 On Noise and Scale in Spatial Partitioning

2.3.2 Clustering Algorithm with Multiple Scale Parameters for Noisy Data

2.3.3 Robust Fuzzy Relational Data Clustering Algorithm

2.3.4 Numerical Experiments

2.4 Partitioning of Spatial Object Data by Unidimensional Scaling 2.4.1 A Note on the Use of Unidimensional Scaling

2.4.2 Basic Principle of Unidimensional Scaling in Data Clustering

2.4.3 Analysis of Simulated Data

2.4.4 UDS Clustering of Remotely Sensed Data

2.5 Unraveling Spatial Objects with Arbitrary Shapes Through Mixture Decomposition Clustering 2.5.1 On Noise and Mixture Distributions in Spatial Data

2.5.2 A Remark on the Mining of Spatial Features with Arbitrary Shapes

2.5.3 A Spatial-Feature Mining Model (RFMM) Based on Regression- Class Mixture Decomposition ( RCMD)

2.5.4 The RFMM with Genetic Algorithm (RFMM-GA)

2.5.5 Applications of RFMM-GA in the Mining of Features in Remotely Sensed Images

2.6 Cluster Characterization by the Concept of Convex Hull 2.6.1 A Note on Convex Hull and its Computation

2.6.2 Basics of the Convex Hull Computing Neural Network ( CHCNN) Model

2.6.3 The CHCNN Architecture

2.6.4 Applications in Cluster Characterization

Statistical Approach to the Identification of Separation Surface for Spatial Data

3.1 A Brief Background About Statistical Classification

3.2 The Bayesian Approach to Data Classification

3.2.1 A Brief Description of Bayesian Classification Theory

3.2.2 Naive Bayes Method and Feature Selection in Data Classification

3.2.3 The Application of Nai AE ve Bayes Discriminant Analysis in Client Segmentation for Product Marketing

3.2.4 Robust Bayesian Classification Model

3.3 Mixture Discriminant Analysis 3.3.1 A Brief Statement About Mixture Discriminant Analysis

3.3.2 Mixture Discriminant Analysis by Optimal Scoring

3.3.3 Analysis Results and Interpretations

3.4 The Logistic Model for Data Classification 3.4.1 A Brief Note About Using Logistic Regression as a Classifier

3.4.2 Data Manipulation for Client Segmentation

3.4.3 Logistic Regression Models and Strategies for Credit Card Promotion

3.4.4 Model Comparisons and Validations

3.5 Support Vector Machine for Spatial Classification 3.5.1 Support Vector Machine as a Classifier

3.5.2 Basics of Support Vector Machine

3.5.3 Experiments on Feature Extraction and Classification by SVM

Algorithmic Approach to the Identification of Classification Rules or Separation Surface for Spatial Data

4.1 A Brief Background About Algorithmic Classification

4.2 The Classification Tree Approach to the Discovery of Classification Rules in Data 4.2.1 A Brief Description of Classification and Regression tree ( CART)

4.2.2 Client Segmentation by CART

4.3 The Neural Network Approach to the Classification of Spatial Data 4.3.1 On the Use of Neural Networks in Spatial Classification

4.3.2 The Knowledge-Integrated Radial Basis Function (RBF) Model for Spatial Classification

4.3.3 An Elliptical Basis Function Network for Spatial Classification

4.4 Genetic Algorithms for Fuzzy Spatial Classification Systems 4.4.1 A Brief Note on Using GA to Discover Fuzzy Classification Rules

4.4.2 A General Framework of the Fuzzy Classification System

4.4.3 Fuzzy Rule Acquisition by GANGO

4.4.4 An Application in the Classification of Remote Sensing Data

4.5 The Rough Set Approach to the Discovery of Classification Rules in Spatial Data 4.5.1 Basic Ideas of the Rough Set Methodology for Knowledge Discovery

4.5.2 Basic Notions Related to Spatial Information Systems and Rough Sets

4.5.3 Interval-Valued Information Systems and Data Transformation

4.5.4 Knowledge Discovery in Interval-Valued Information Systems

4.5.5 Discovery of Classification Rules for Remotely Sensed Data

4.5.6 Classification of Tree Species with Hyperspectral Data

4.6 A Vision-Based Approach to Spatial Classification 4.6.1 On Scale and Noise in Spatial Data Classification

4.6.2 The Vision-Based Classification Method

4.6.3 Experimental Results

4.7 A Remark on the Choice of Classifiers

Discovery of Spatial Relationships in Spatial Data

5.1 On Mining Spatial Relationships in Spatial Data

5.2 Discovery of Local Patterns of Spatial Association 5.2.1 On the Measure of Local Variations of Spatial Associations

5.2.2 Local Statistics and their Expressions as a Ratio of Quadratic Forms

5.3 Dicovery of Spatial Non-Stationarity Based on the Geographically Weighted Regression Model 5.3.1 On Modeling Spatial Non- Stationarity within the Parameter- Varying Regression Framework

5.3.2 Geographically Weighted Regression and the Local–Global Issue About Spatial Non- Stationarity

5.3.3 Local Variations of Regional Industrialization in Jiangsu Province, P. R. China

5.3.4 Discovering Spatial Pattern of Influence of Extreme Temperatures on Mean Temperatures in China

5.4 Testing for Spatial Autocorrelation in Geographically Weighted Regression

5.5 A Note on the Extentions of the GWR Model

5.6 Discovery of Spatial Non-Stationarity Based on the Regression- Class Mixture Decomposition Method 5.6.1 On Mixture Modeling of Spatial Non- Stationarity in a Noisy Environment

5.6.2 The Notion of a Regression Class

5.6.3 The Discovery of Regression Classes under Noise Contamination

5.6.4 The Regression-Class Mixture Decomposition (RCMD) Method for knowledge Discovery in Mixed Distribution

5.6.5 Numerical Results and Observations

5.6.6 Comments About the RCMD Method

5.6.7 A Remote Sensing Application

5.6.8 An Overall View about the RCMD Method

Discovery of Structures and Processes in Temporal Data

6.1 A Note on the Discovery of Generating Structures or Processes of Time Series Data

6.2 The Wavelet Approach to the Mining of Scaling Phenomena in Time Series Data 6.2.1 A Brief Note on Wavelet Transform

6.2.2 Basic Notions of Wavelet Analysis

6.2.3 Wavelet Transforms in High Dimensions

6.2.4 Other Data Mining Tasks by Wavelet Transforms

6.2.5 Wavelet Analysis of Runoff Changes in the Middle and Upper Reaches of the Yellow River in China

6.2.6 Wavelet Analysis of Runoff Changes of the Yangtze River Basin

6.3 Discovery of Generating Structures of Temporal Data with Long- Range Dependence 6.3.1 A Brief Note on Multiple Scaling and Intermittency of Temporal Data

6.3.2 Multifractal Approach to the Identification of Intermittency in Time Series Data

6.3.3 Experimental Study on Intermittency of Air Quality Data Series

6.4 Finding the Measure Representation of Time Series with Intermittency 6.4.1 Multiplicative Cascade as a Characterization of the Time Series Data

6.4.2 Experimental Results

6.5 Discovery of Spatial Variability in Time Series Data 6.5.1 Multifractal Analysis of Spatial Variability Over Time

6.5.2 Detection of Spatial Variability of Rainfall Intensity

6.6 Identification of Multifractality and Spatio-Temperal Long Range Dependence in Multiscaling Remote Sensing 6.6.1 A Note on Multifractality and Long- Range Dependence in Remote Sensing Data

6.6.2 A Proposed Methodology for the Analysis of Multifractality and Long- Range Dependence in Remote Sensing Data

6.7 A Note on the Effect of Trends on the Scaling Behavior of Time Series with Long- Range Dependence

Summary and Outlooks

7.1 Summary

7.2 Directions for Further Research 7.2.1 Discovery of Hierarchical Knowledge Structure from Relational Spatial Data

7.2.2 Errors in Spatial Knowledge Discovery

7.2.3 Other Challenges

7.3 Concluding Remark

Bibliography

Author Index

Subject Index