Representation of Space and Time in Demand Forecasting Models

Problem

Typically, space and time are both represented in a discrete manner in aggregate trip/tour-based models as well as most activity-based models.  Space is represents by a collection of non-overlapping zones; where as time, in most models, is represented by discrete time-of-day periods.  Yet the continuous nature of the spatial and temporal domains raises the important issue concerning the level of detail to be incorporated in travel demand models.  This is because aggregation over space and time leads to loss of information and various aggregation biases.  The examples of practical importance include modeling transit access (extremely sensitive to details of walk distance and pedestrian friendliness) and congestion pricing (where the shift from the peak hour to shoulders is frequently in the focus of analysis).  Yet, the reduction of the level of detail is often necessary in order to uncover pattern in robust statistical terms and to yield better model forecasting capability as well as to ensure reasonable runtime in the model system application at the regional level. 

To-date, the relationships between spatial and temporal resolution and model sensitivity and accuracy are not at all well understood.  Do smaller zones and more time-of-day periods always yield ‘better’ travel demand models in practical terms?  Should models be unaffected by the choices of spatial and temporal analysis units – if this is at all possible?  Or, should spatial and time be represented in a continuous manner to avoid any possible dependence on the level of resolution?

Objective

This research project seeks to answers the above questions that are fundamental to model development and application.  It is intended to outline the fundamental research avenues as well as formulate practical recommendations for incorporation of the research results in regional travel models.  It will also seek to evaluate the compromises involved in terms of data requirements and model accuracy.

Key Words

Spatial scale, temporal scale, aggregation bias, traffic analysis zones

Related Work

The representation of space and time in urban models is being explored in the fields of geography and regional sciences.  Yet the discussion is the context of demand models is relatively limited and scattered.  As part of this research, a synthesis is necessary to survey the relevant literature on this topic.

Urgency/Priority

With greater emphasis on micro-scale projects (e.g. transit oriented development) and increasing availability of high-resolution spatial data, many regions are grappling with questions of spatial detail in their current models and future model designs. This study may be viewed as a high priority research item. 

Cost

$ 140,000

User Community

FTA, AASHTO, FHWA, MPOs

Implementation

The findings from this research would serve as guidelines for MPOs in determining the spatial and temporal resolution in their demand forecasting models.

Effectiveness

This research will inform model developers the trade-offs between data needs and model complexity, on the one hand, and forecasting capability and accuracy, on the other, thereby leading to effective use of model development resources.  



Sponsoring Committee: ADB40, Transportation Demand Forecasting
Source Info: Committee members
Date Posted: 08/15/2007
Date Modified: 08/15/2007
Index Terms: Spatial analysis, Traffic distribution, Traffic analysis, Temporal analysis, Aggregation levels, Travel demand, Travel demand management,

 
Subjects    
Highways, Planning and Forecasting

Please click here if you wish to share information or are aware of any research underway that addresses issues in this research needs statement. The information may be helpful to the sponsoring committee in keeping the statement up-to-date.


Integrating Demand and Supply Travel Simulation Models

Problem

In research and practice, there have been many advances on both the travel demand and supply side simulation modeling, but there has been less attention to the integration of these advanced models.  Current integration techniques (i.e. loose coupling of these models to transfer data from one to the other) do not take advantage of either the added behavioral data available in each model nor do they achieve consistency among parameters that are used in both demand and supply models.

Objective

Research to explore the various different levels of integration that are possible for demand and supply travel simulation models, which may include the following:

  • Incorporating specific features of demand modeling into supply side modeling, such as path-building parameters;
  • Incorporating specific features of supply modeling into demand side modeling, such as value of time parameters for different market segments (i.e. truck trips, work trips, by income group, etc.) into demand model components, such as trip distribution and time of day;
  • Incorporating specific features of activity-based simulation modeling in advanced travel simulation models, such as dynamic traffic assignment, route choice or traffic microsimulation models, and vice versa; and
  • Integrating demand and supply travel simulation models to take advantage of behavioral details of each person and each vehicle developed in the demand model for use in refining the supply model (i.e. value of time for each person could be used in assigning trips to toll roads in the traffic operation or route choice model).

Key Words

Travel demand modeling, transportation supply modeling, model integration

Related Work

Special sessions on this topic were organized for the INFORMS annual meeting in November 2006, the TRB annual meeting in January 2007, and the 11th World Conference on Transport Research in June 2007.  Studies presented at these sessions are representative of the current state-of-the-art and practice in this topic area.  

Urgency/Priority

This research topic is relevant to the enhancement of the traditional trip-based models as well as the activity-based models.  It is considered a medium to high priority.

Cost

$150,000

User Community

FHWA, TMIP,MPOs

Implementation

The methodologies developed in this research would be incorporated in current travel forecasting models.

Effectiveness

Improved travel forecasting models resulting from this research would serve as better tools to support transportation investment decision-making.



Sponsoring Committee: ADB40, Transportation Demand Forecasting
Source Info: Committee members
Date Posted: 08/15/2007
Date Modified: 08/15/2007
Index Terms: Travel demand management, Travel demand, Supply, Travel simulators, Traffic assignment, Route choice,

 
Subjects    
Highways, Public Transportation, Planning and Forecasting, Economics

Please click here if you wish to share information or are aware of any research underway that addresses issues in this research needs statement. The information may be helpful to the sponsoring committee in keeping the statement up-to-date.


Policy Sensitivity: Trip-Based vs. Tour- and Activity-Based Models

Problem
 
The types of questions being asked of travel demand models are expanding well beyond mere traffic projections and transit ridership. Decision-makers are now concerned about the technical assessment of the effectiveness of pricing techniques, transit-oriented development, greenhouse gas consequences, peak spreading, and detailed equity and environmental justice analyses.
 
The activity-based modeling approach has emerged as a potentially more suitable tool for answering these types of questions than the traditional trip-based approach. When assessing the relative merits of these two types of modeling approaches, one must make the assessment based on not only how well the models replicate observed travel conditions, but also the models’ sensitivity to different demand management policies and strategies. Yet, the policy sensitivities of activity-based models have not been fully vetted.  This research will evaluate the relative effectiveness of trip- and activity-based models at answering these policy questions.
 
Objective
 
This research item will be geared toward comparing sensitivities of trip and activity-based models to the following key sensitivity areas:
 
  • Price sensitivity, such as in tolling or congestion-pricing applications
  • Sensitivity to transit-oriented development patterns and the resulting transit ridership consequences
  • FTA New Starts ridership and user-benefit applicability
  • Equity and environmental justice sensitivities, such as effects of transportation and development scenarios on low-income or transit-dependent households
 
Key Words
 
Travel demand models, sensitivity analysis, demand management policies and strategies
 
Related Work
 
Columbus, New York, San Francisco, and Sacramento are notable examples of cities that have adopted tour-based models while other cities, including Sacramento, Denver, and Atlanta, are in the process of developing these models. Some of these cities have (or will have) both trip- and tour-based models in operation that could be well-suited for comparisons with cooperation of the relevant planning agencies.
 
Urgency/Priority
 
Given the relatively slow uptake of activity based models, identifying the strengths and weaknesses of these newer models relative to traditional models is relevant for a large portion of MPOs and planning agencies. This research problem is regarded as a high priority.
 
Cost
 
$140,000
 
User Community
 
FHWA, FTA, ASSHTO, TMIP, MPOs
 
Implementation
 
The findings obtained from this research can be used by MPOs and DOTs across the country to make decision regarding the development and updating of their travel demand models.
 
Effectiveness
 
This research will rely heavily on existing trip-based and activity-based model systems in the United States. By studying the aforementioned policy sensitivities across a repre­sentative sample of models from across the US, this research is expected to serve as a guidance tool for practitioners. Clearly, the effectiveness of this research will depend on the willingness of the planning agencies to share their models for the purpose of this study.


Sponsoring Committee: ADB40, Transportation Demand Forecasting
Source Info: Committee members, Innovations in Travel Modeling 2006 conference
Date Posted: 08/15/2007
Date Modified: 08/15/2007
Index Terms: Travel demand, Travel demand management, Congestion pricing, Road pricing, Transit oriented development, Greenhouse gases, Environmental justice,

 
Subjects    
Highways, Public Transportation, Planning and Forecasting, Finance

Please click here if you wish to share information or are aware of any research underway that addresses issues in this research needs statement. The information may be helpful to the sponsoring committee in keeping the statement up-to-date.


Activity- and Tour-Based Models in Practice: Optimizing Model Performance

Problem
 
While the research community has already demonstrated the theoretical merits of activity and tour-based models, the practitioners are only gradually warming up to these model types as potential policy tools. One of the recurrent themes and a major concern in all discussions about activity and tour-based models is the computational resources and time required to execute these models. A key step in helping practitioners judge whether or not to make the transition to activity and tour based models is to explore ways of optimizing model performance.
 
Objective
 
This research item will focus on (a) assessing the computational challenges arisen from executing existing activity- and tour-based models and (b) identifying possible strategies for overcoming these challenges.  The research will also evaluate the potential tradeoffs between model accuracy and computational efficiency improvements resulting from these strategies based on the following performance measures:
 
  • Sample size of population used for simulation
  • Model run time
  • Model convergence measures
  • Accuracy of predicted spatial distribution of trips
  • Accuracy of predicted temporal distribution of trips
  • Accuracy of predicted traffic patterns on major highway and transit routes
 
Key Words
 
Travel demand forecasting, model run times, model accuracy, computational efficiency
 
Related Work
 
Columbus, New York and San Francisco are notable examples of cities that have adopted tour-based models while other cities, including Sacramento, Denver, and Atlanta, are in the process of developing these models. Other metropolitan areas, such as Dallas/Fort-Worth, have also been used to demonstrate the activity-based modeling approach. Experiences from these regions will be valuable in identifying the tradeoffs between model accuracy and computational efficiency.
 
Urgency/Priority
 
This research is considered a priority as the research will help moving activity- and tour-based models into practice by reducing the computational burdens typically associated with the application of these models. Given the fact that these models are gaining the interests of many MPOs, this research will also help guide the MPOs in choosing the suitable activity- and/or tour-based model structure to support their planning activities.  The priority of this research is medium to high.
 
Cost
 
$100,000
 
User Community
 
FHWA, TMIP, MPOs
 
Implementation
 
Recommendations derived from this research can be used to fine-tune existing activity and tour-based models to achieve optimum performance for the desired level of model accuracy. They should also be used to guide future development of activity- and tour-based models.
 
Effectiveness
 
This research will rely heavily on existing activity-based model systems in the United States. By studying the aforementioned performance measures across a representative sample of models from across the US, this research is expected to serve as a guidance tool for practitioners. Clearly, the effectiveness of this research will depend on the willingness of the planning agencies to share their models for the purpose of this study.


Sponsoring Committee: ADB40, Transportation Demand Forecasting
Source Info: Committee members
Date Posted: 08/15/2007
Date Modified: 08/15/2007
Index Terms: Travel demand, Travel demand management, Computational efficiency, Run time (Computers), Accuracy,

 
Subjects    
Highways, Public Transportation, Planning and Forecasting

Please click here if you wish to share information or are aware of any research underway that addresses issues in this research needs statement. The information may be helpful to the sponsoring committee in keeping the statement up-to-date.