Title Page

ABSTRACT

Contents

1. INTRODUCTION 11

1.1. History of waste management 11

1.2. Environmental policies 16

1.3. Research question and objective 18

1.4. Statistical analysis and Predictive modelling 18

1.5. Scope and area of study 28

1.6. Related work 29

1.7. Purpose, relevance and contribution of the study 30

2. THEORETICAL FRAMEWORK 31

2.1. Review of basic concepts 31

2.2. Previous works related to the study of policy impact 32

2.3. Previous works related to the study of environmental policy's impact 32

2.4. Previous works related to the study of environmental systems through predictive modeling 33

2.5. Previous works that study the application of predictive modeling for policy impact 34

2.6. Previous works that study the application of predictive modeling for environmental systems 35

2.7. Previous works that study factors affecting CO2 emissions 35

2.8. Previous works that apply GWR technique for environmental impact investigation 46

2.9. Previous works that apply MGWR technique for environmental impact investigation 50

2.10. Advantages and limitations of GWR 53

3. METHODOLOGY 53

3.1. Research design and approach 53

3.2. Scope and area of study 54

3.3. Methods 56

3.4. Description of variables 59

3.4.1. Carbon Dioxide equivalent 59

3.4.2. Socioeconomic variables 60

3.4.3. Energy consumption and expenditure per sector 63

3.4.4. Land Cover 65

3.4.5. Hazardous waste 66

3.5. Data analysis and preprocessing 68

3.5.1. Data preparation and preliminary analysis 68

3.5.2. Spatial Interpolation of Total Releases 68

3.5.3. GWR Analysis 69

3.5.4. MGWR Analysis 69

4. RESULTS 71

4.1. Data preparation and preliminary analysis 71

4.1.1. Data preparation 71

4.1.2. GMW and MGWR analysis 71

4.2. Variable selection 72

4.2.1. Correlation Analysis 72

4.2.2. Stepwise Regression 72

4.2.3. Validation 73

4.2.4. Ensuring Robustness and Interpretability and Application in Environmental Science 73

4.3. Performance statistics 75

4.4. Characteristics of spatial distribution 76

4.5. Spatial heterogeneity of influencing factors of CO2 77

4.5.1. GWR 78

4.5.2. GWR Spatial distribution of regression coefficients by regions 84

4.5.3. MGWR 86

4.5.4. MGWR Spatial distribution of regression coefficients by regions 93

4.5.5. SHAP plot 94

5. DISCUSSION 95

5.1. Poverty and race features 95

5.2. CO2eq distribution 98

5.3. Spatial analysis of predictors and locations 100

5.3.1. GWR spatial distribution of regression coefficients by regions 103

5.3.2. MGWR spatial distribution of regression coefficients by regions 107

5.4. Big cities vs. small communities 108

5.5. Policy implications 113

6. CONCLUSION 115

REFERENCES 117

국문초록 141

Table 1.1. Advantages and disadvantages of predictive modelling classes 25

Table 1.2. Advantages and disadvantages of different types of regression models 25

Table 2.1. Comparison of studies focusing on determining driving factors behind CO2 emissions 37

Table 2.2. Comparison of studies applying GWR to determine driving factors behind CO2 emissions 47

Table 2.3. Comparison of studies applying MGWR to determine driving factors behind CO2 emissions 51

Table 3.1. Summary of the parts contained in the title 40 CFR, of the Subtitle C of the RCRA 55

Table 3.2. Description of parameters for the variable CO2eq 60

Table 3.3. Description of parameters for the socioeconomic variables 60

Table 3.4. Description of parameters for the energy consumption and expenditure variable 64

Table 3.5. Description of parameters for the land cover variable 66

Table 3.6. Description of parameters for the hazardous waste variable 67

Table 4.1. Description of variables 74

Table 4.2. Summary of performance metrics for GWR and MGWR models 76

Table 4.3. Statistics summary for CO2eq emissions 77

Table 4.4. Location of ZIP codes with highest and lowest mean CO2eq emissions 77

Table 4.5. Statistical description of variables for GWR model 81

Table 4.6. GWR regression results 82

Table 4.7. Statistical description of variables for MGWR model 90

Table 4.8. MGWR regression results 91

Table 5.1. Regression coefficients for significant features in extreme end ZIP codes 108

Table 5.2. Relationship between variables and policies to address them 113

Figure 1.1. Waste management hierarchy 18

Figure 1.2. Route map for GWR technique selection 23

Figure 1.3. Map for choosing the right predictive modelling technique 24

Figure 3.1. Evolution of significant RCRA legislation 55

Figure 3.2. Framework of the methodology 70

Figure 4.1. Distribution of CO2eq emissions at ZIP code level 77

Figure 4.2. Distribution of R2 for GWR model 78

Figure 4.3. Spatial distribution of regression coefficients for GWR model 78

Figure 4.4. Spatial distribution of regression coefficients for MGWR model 86

Figure 4.5. SHAP plot for MGWR model 94