Date of Completion

5-12-2019

Embargo Period

10-20-2019

Advisors

Karthik C. Konduri, Nicholas E. Lownes, Jin Zhu

Field of Study

Civil Engineering

Degree

Master of Science

Open Access

Open Access

Abstract

In order to apply microsimulation-based models of land use and travel demand, socio-economic and demographic attributes are required for each individual in a region. This disaggregate level information is not readily available and planners resort to population synthesis procedures. This research includes two studies that are focused on developing alternative paradigms for population synthesis and for estimating sample household weights. In the first study, a simulation-based technique for multi-level population synthesis using a Hidden Markov Model (HMM) framework is presented. A comparative analysis is carried out to highlight the feasibility and applicability of the proposed approach in generating consistent multilevel agents while adhering to geography-based controls and heterogeneity. As part of the second study, an analytical procedure for estimating sample household weights is proposed that helps estimate consistent weights using disaggregate information with sparse attribute categories by controlling both at the household and person level. Different configurations of the system of linear equations are formulated and evaluated for various sets of block groups as the geographical units. Finally, the synthetic population is generated for ten block groups in Connecticut using the proposed synthesizing framework and weight estimation procedure. The analysis of synthetic outputs confirms that the proposed weight estimation procedure is comparable with the heuristic-based approaches and can be used as an alternate weight estimation routine for simulating more consistent household and person level attributes or drawing households from the sample to obtain a synthetic population that closely match the available aggregate information.

Major Advisor

Karthik C. Konduri

Available for download on Sunday, October 20, 2019

Share

COinS