Mahmoud Javanmardi

Dynamic Activity Generation Model Using Competing Hazard Formulation

Rule-based activity scheduling microsimulation models often generate activities for individuals to engage in randomly, on the basis of observed activity rates from survey data. These microsimulation models try to represent more closely the process of activity pattern development. However, the dynamics underlying the activity generation process are often not considered, especially in regard to competition between activities for the limited time resource. This work, then, develops a methodology for generating activities on the basis of the time since the last activity of the same type was generated, by using a hazard-based formulation. In addition, the model explicitly accounts for the competition between activities through the use of a competing hazard framework. The results show that observed activity rates and temporal distributions from survey data can be replicated through simulation of the model in an activity-based scheduling model, the agent-based dynamic activity planning and travel scheduling (ADAPTS) activity scheduler.

Resolution of Activity Scheduling Conflicts: Reverse Pairwise Comparison of In-Home and Out-of-Home Activities

This paper proposes a new methodology for solving a pairwise comparison problem from a polychotomous classification. The proposed methodology reverses the pairwise coupling and employs the Bradley–Terry formulation for estimating pairwise comparison models. This methodology was implemented through relevant models in Agent-Based Dynamic Activity Planning and Scheduling (ADAPTS) and aims for solving pairwise activity conflicts. At one step of the framework, given the default algorithm, agents have to make a choice between in-home and out-of-home activities. At this time step, the reverse pairwise comparison models are called to investigate the agents’ resources and conditions for making the most probable decision about which activity to engage in. This work was conducted in conjunction with an earlier effort to incorporate in-home activities into ADAPTS, and the models proposed provide a behavioral decision-making process for this purpose.

Choice Set Imputation: Two-Step Weighted Stratified and Hazard-Based Approach

The problem of choice set formation for decision makers is an important subject in discrete choice modeling, especially when the choice set contains a large number of elemental alternatives. In general, the choice set of an individual could be a randomly sampled choice set; however, this is claimed to be a behaviorally unacceptable practice because of the fallacious assumption of individuals’ full knowledge of potential random choices. This issue brings up the need to devise methods to logically allocate credible choice alternatives for individuals. Although the use of these methods could be dependent on specific applications, this study attempted to identify the distinction between model estimation and prediction steps in the context of residential location choice modeling. From a theoretical point of view, the paper proposes a modified weighted stratified sampling approach that is an improved version of random sampling for model estimation. The approach is believed to be a better replicate of the universal choice set than other sampling methods, and it is capable of resulting in consistent estimates even with small sample sizes. The estimated model was applied in a simulation framework with a hazard-based imputed choice set approach for prediction.

Household Travel Data Simulation Tool: Software and Its Applications for Impact Analysis

Data transferability is seen as an alternative solution to costly travel surveys for urban areas where regular travel data are difficult to collect, especially in small and mid-sized communities. A comprehensive travel data transferability model and a software tool that can facilitate travel data transferability and simulate synthetic household-level disaggregate travel data have been developed. The model is built on earlier transferability studies by a significant enhancement of the approach and resolution of many limitations of previous studies. The software tool has been tested on two case studies in Des Moines, Iowa, and New York State. Nine household-level travel attributes are simulated for the synthetic population of these regions. A comparison of the simulated travel data with the actual observed data, obtained from the National Household Travel Survey add-on samples, proves the accuracy of the model. It is also shown that updating the parameters of the distributions of travel attributes can further improve the results. The model is then used for some basic policy evaluations and a sensitivity analysis that includes scenarios such as changes to the demographics, aging population, and investments in the education system. The result of the sensitivity analysis also confirmed the wide capabilities of the model.

Evaluation Methods for Estimating Vehicle Miles Traveled with GPS Travel Survey Data

This research investigated the degree to which traditional routing algorithms, including those that took congestion levels into consideration, could be used to accurately predict GPS-recorded vehicle miles traveled (VMT) if only activity locations were known. Given recent policy interest in distance-based charges, the ability to predict household VMT accurately is an important research area because it can improve the quality of distributional assessments of distance-based proposals. This analysis found that shortest-time travel paths that incorporated congestion levels performed best across all income groups, urban locations, and trip lengths when compared with shortest-travel time paths with no congestion or shortest-distance paths. The average margin of error from this analysis was considerably smaller than those found in other studies. Failure to incorporate congestion effects into distance estimates consistently resulted in underestimation of household travel distance, sometimes rather significantly.