[PDF]
Next:
Abstract
Contents
Synthesizing Agents and Relationships for Land Use / Transportation Modelling
David R. Pritchard
Contents
Abstract
Acknowledgments
List of Tables
List of Figures
1.
Introduction
2.
Previous Work
2.1
The ILUTE Model
2.2
Mathematics for Fitting Contingency Tables
2.2.1
Notation
2.2.2
History and Properties of Iterative Proportional Fitting
2.2.3
Generalizations of the IPF Method
2.2.4
Log-Linear Models
2.2.5
IPF and Log-Linear Models
2.2.6
Zero Cells
2.3
Population Synthesis
2.3.1
Zone-by-Zone IPF
2.3.2
Multizone IPF
2.3.3
Synthesis Examples Using IPF
2.3.4
Reweighting and Combinatorial Optimization
3.
Data Sources and Definitions
3.1
Family and Household Definitions
3.2
Agent Attributes
3.3
Exploration of a Summary Table
4.
Method Improvements
4.1
Simplifying the PUMS
4.2
Sparse List-Based Data Structure
4.2.1
Algorithmic Details
4.2.2
Discussion
4.3
Fitting to Randomly Rounded Margins
4.3.1
Modified Termination Criterion
4.3.2
Hierarchical Margins
4.3.3
Projecting onto Feasible Range
4.4
Synthesizing Agent Relationships
4.4.1
Fitting Populations Together
4.4.2
Conditioned Monte Carlo
4.4.3
Summary
5.
Implementation
5.1
Population Universe
5.2
Relationship Model
5.3
Attributes
5.4
Shared Attribute Selection
5.4.1
Households and Dwellings
5.4.2
Families and Persons
5.4.3
Households/Dwellings and Families
5.4.4
Households and Non-Family Persons
5.5
Software
5.5.1
IPF Implementation
5.5.2
Random Rounding and Area Suppression
5.5.3
Conditional Monte Carlo
5.6
Results
6.
Evaluation
6.1
Goodness-of-Fit Measures
6.2
Tests of IPF Method and Input Margins
6.2.1
Source Sample
6.2.2
1D Margins versus 2-3D Margins
6.2.3
Zone-by-zone versus Multizone
6.3
Effects of Random Rounding
6.4
Effects of Monte Carlo
7.
Conclusion
Bibliography
A. Attribute Definitions
A.1
Person Attributes
A.2
Family Attributes
A.3
Dwelling/Household Attributes
B. Detailed Results
Next:
Abstract
Contents
