Spatial Big Data Science (eBook)

Preface 6
Acknowledgements 7
Contents 8
Acronyms 11
Part I Overview of Spatial Big Data Science 12
1 Spatial Big Data 13
1.1 What Is Spatial Big Data? 13
1.2 Societal Applications 16
1.3 Challenges 18
1.3.1 Implicit Spatial Relationships 18
1.3.2 Spatial Autocorrelation 19
1.3.3 Spatial Anisotropy 19
1.3.4 Spatial Heterogeneity 20
1.3.5 Multiple Scales and Resolutions 20
1.4 Organization of the Book 21
References 23
2 Spatial and Spatiotemporal Big Data Science 24
2.1 Input: Spatial and Spatiotemporal Data 25
2.1.1 Types of Spatial and Spatiotemporal Data 25
2.1.2 Data Attributes and Relationships 26
2.2 Statistical Foundations 27
2.2.1 Spatial Statistics for Different Types of Spatial Data 27
2.2.2 Spatiotemporal Statistics 29
2.3 Output Pattern Families 30
2.3.1 Spatial and Spatiotemporal Outlier Detection 30
2.3.2 Spatial and Spatiotemporal Associations, Tele-Connections 31
2.3.3 Spatial and Spatiotemporal Prediction 33
2.3.4 Spatial and Spatiotemporal Partitioning (Clustering) and Summarization 38
2.3.5 Spatial and Spatiotemporal Hotspot Detection 41
2.3.6 Spatiotemporal Change 43
2.4 Research Trend and Future Research Needs 44
2.5 Summary 46
References 46
Part II Classification of Earth Observation Imagery Big Data 54
3 Overview of Earth Imagery Classification 55
3.1 Earth Observation Imagery Big Data 55
3.2 Societal Applications 56
3.3 Earth Imagery Classification Algorithms 58
3.4 Generating Derived Features (Indices) 60
3.5 Remaining Computational Challenges 61
References 63
4 Spatial Information Gain-Based Spatial Decision Tree 65
4.1 Introduction 65
4.1.1 Societal Application 65
4.1.2 Challenges 67
4.1.3 Related Work Summary 68
4.2 Problem Formulation 68
4.3 Proposed Approach 71
4.3.1 Basic Concepts 71
4.3.2 Spatial Decision Tree Learning Algorithm 76
4.3.3 An Example Execution Trace 77
4.4 Evaluation 79
4.4.1 Dataset and Settings 79
4.4.2 Does Incorporating Spatial Autocorrelation Improve Classification Accuracy? 81
4.4.3 Does Incorporating Spatial Autocorrelation Reduce Salt-and-Pepper Noise? 81
4.4.4 How May One Choose ?, the Balancing Parameter for SIG Interestingness Measure? 82
4.5 Summary 83
References 84
5 Focal-Test-Based Spatial Decision Tree 85
5.1 Introduction 85
5.2 Basic Concepts and Problem Formulation 88
5.2.1 Basic Concepts 88
5.2.2 Problem Definition 91
5.3 FTSDT Learning Algorithms 91
5.3.1 Training Phase 92
5.3.2 Prediction Phase 96
5.4 Computational Optimization: A Refined Algorithm 97
5.4.1 Computational Bottleneck Analysis 97
5.4.2 A Refined Algorithm 98
5.4.3 Theoretical Analysis 101
5.5 Experimental Evaluation 103
5.5.1 Experiment Setup 103
5.5.2 Classification Performance 104
5.5.3 Computational Performance 106
5.6 Discussion 110
5.7 Summary 111
References 111
6 Spatial Ensemble Learning 113
6.1 Introduction 113
6.2 Problem Statement 115
6.2.1 Basic Concepts 115
6.2.2 Problem Definition 119
6.3 Proposed Approach 120
6.3.1 Preprocessing: Homogeneous Patches 120
6.3.2 Approximate Per Zone Class Ambiguity 122
6.3.3 Group Homogeneous Patches into Zones 123
6.3.4 Theoretical Analysis 124
6.4 Experimental Evaluation 126
6.4.1 Experiment Setup 126
6.4.2 Classification Performance Comparison 127
6.4.3 Effect of Adding Spatial Coordinate Features 129
6.4.4 Case Studies 130
6.5 Summary 132
References 133
Part III Future Research Needs 135
7 Future Research Needs 136
7.1 Future Research Needs 136
7.2 Summary 138
Reference 138