Matching Items (3)

153860-Thumbnail Image.png

Threshold regression estimation via lasso, elastic-net, and lad-lasso: a simulation study with applications to urban traffic data

Description

Threshold regression is used to model regime switching dynamics where the effects of the explanatory variables in predicting the response variable depend on whether a certain threshold has been crossed. When regime-switching dynamics are present, new estimation problems arise related

Threshold regression is used to model regime switching dynamics where the effects of the explanatory variables in predicting the response variable depend on whether a certain threshold has been crossed. When regime-switching dynamics are present, new estimation problems arise related to estimating the value of the threshold. Conventional methods utilize an iterative search procedure, seeking to minimize the sum of squares criterion. However, when unnecessary variables are included in the model or certain variables drop out of the model depending on the regime, this method may have high variability. This paper proposes Lasso-type methods as an alternative to ordinary least squares. By incorporating an L_{1} penalty term, Lasso methods perform variable selection, thus potentially reducing some of the variance in estimating the threshold parameter. This paper discusses the results of a study in which two different underlying model structures were simulated. The first is a regression model with correlated predictors, whereas the second is a self-exciting threshold autoregressive model. Finally the proposed Lasso-type methods are compared to conventional methods in an application to urban traffic data.

Contributors

Agent

Created

Date Created
2015

149519-Thumbnail Image.png

An assessment of stochastic variability and convergence characteristics in travel microsimulation models

Description

In the middle of the 20th century in the United States, transportation and infrastructure development became a priority on the national agenda, instigating the development of mathematical models that would predict transportation network performance. Approximately 40 years later, transportation planning

In the middle of the 20th century in the United States, transportation and infrastructure development became a priority on the national agenda, instigating the development of mathematical models that would predict transportation network performance. Approximately 40 years later, transportation planning models again became a national priority, this time instigating the development of highly disaggregate activity-based traffic models called microsimulations. These models predict the travel on a network at the level of the individual decision-maker, but do so with a large computational complexity and processing time requirement. The vast resources and steep learning curve required to integrate microsimulation models into the general transportation plan have deterred planning agencies from incorporating these tools. By researching the stochastic variability in the results of a microsimulation model with varying random number seeds, this paper evaluates the number of simulation trials necessary, and therefore the computational effort, for a planning agency to reach stable model outcomes. The microsimulation tool used to complete this research is the Transportation Analysis and Simulation System (TRANSIMS). The requirements for initiating a TRANSIMS simulation are described in the paper. Two analysis corridors are chosen in the Metropolitan Phoenix Area, and the roadway performance characteristics volume, vehicle-miles of travel, and vehicle-hours of travel are examined in each corridor under both congested and uncongested conditions. Both congested and uncongested simulations are completed in twenty trials, each with a unique random number seed. Performance measures are averaged for each trial, providing a distribution of average performance measures with which to test the stability of the system. The results of this research show that the variability in outcomes increases with increasing congestion. Although twenty trials are sufficient to achieve stable solutions for the uncongested state, convergence in the congested state is not achieved. These results indicate that a highly congested urban environment requires more than twenty simulation runs for each tested scenario before reaching a solution that can be assumed to be stable. The computational effort needed for this type of analysis is something that transportation planning agencies should take into consideration before beginning a traffic microsimulation program.

Contributors

Agent

Created

Date Created
2010

152703-Thumbnail Image.png

Peak travel in a megacity: exploring the role of infrastructure saturation on the suppression of automobile use

Description

Contrary to many previous travel demand forecasts there is increasing evidence that vehicle travel in developed countries may be peaking. The underlying causes of this peaking are still under much debate and there has been a mobilization of research, largely

Contrary to many previous travel demand forecasts there is increasing evidence that vehicle travel in developed countries may be peaking. The underlying causes of this peaking are still under much debate and there has been a mobilization of research, largely focused at the national scale, to study the explanatory drivers but research focused at the metropolitan scale, where transportation policy and planning are frequently decided, is relatively thin. Additionally, a majority of this research has focused on changes within the activity system without considering the impact transportation infrastructure has on overall travel demand. Using Los Angeles County California, we investigate Peak Car and whether the saturation of automobile infrastructure, in addition to societal and economic factors, may be a suppressing factor. After peaking in 2002, vehicle travel in Los Angeles County in 2010 was estimated at 78 billion and was 20.3 billion shy of projections made in 2002. The extent to which infrastructure saturation may contribute to Peak Car is evaluated by analyzing social and economic factors that may have impacted personal automobile usage over the last decade. This includes changing fuel prices, fuel economy, population growth, increased utilization of alternate transportation modes, changes in driver demographics , travel time and income levels. Summation of all assessed factors reveals there is at least some portion of the 20 billion VMT that is unexplained in all but the worst case scenario. We hypothesize that the unexplained remaining VMT may be explained by infrastructure supply constraints that result in suppression of travel. This finding has impacts on how we see the role of hard infrastructure systems in urban growth and we explore these impacts in the research.

Contributors

Agent

Created

Date Created
2014